Key Takeaways

• Q2 2026 was the second-largest venture quarter on record, with roughly $205 billion raised across 5,000+ startups
• AI companies captured more than 70% of all global venture funding in the quarter
• Anthropic’s $65 billion Series H made it the most valuable private company in the world at $965 billion, passing both OpenAI and SpaceX
• 16 companies raised billion-dollar rounds in Q2, totaling $108.6 billion, or 53% of all funding for the quarter
• The money spread out from pure model labs into defense (Anduril), robotics (NEURA), biotech (Isomorphic), data centers (DayOne), and space (ICEYE)
• China showed up in force, with DeepSeek raising $7.4 billion and Moonshot AI raising $2 billion
• Inference infrastructure became the new battleground, with Baseten and Fireworks AI each raising $1.5 billion within weeks of each other
• Q2 was also the biggest quarter ever for billion-dollar acquisitions, headlined by SpaceX buying Cursor maker Anysphere for $60 billion

1. Anthropic

Total Funding: ~$132B+ | Latest Valuation: $965B | Round: $65B Series H (May 28, 2026)

Anthropic closed a $65 billion Series H on May 28 at a $965 billion post-money valuation, passing OpenAI to become the most valuable private company in the world. The round was co-led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, with each lead reportedly putting in more than $2 billion. Capital Group, Coatue, D1 Capital, GIC, ICONIQ, and XN joined as additional co-leads.

A big part of the story was who else showed up. Memory and compute suppliers Samsung, SK Hynix, and Micron all invested, which reads less like a financial bet and more like a supply chain lining up behind the hardware Anthropic needs to keep scaling. The round also folded in about $15 billion of previously committed hyperscaler money, including $5 billion from Amazon announced in April.

The company said its revenue run-rate had passed $47 billion by early May. Days after the round, Anthropic confidentially filed for an IPO, setting up a possible fall listing that could beat OpenAI to the public market. A Series H is rare in venture history to begin with. The short list of companies that ever reached one includes Facebook, Lyft, and Slack.

Key Investors: Altimeter Capital, Dragoneer, Greenoaks, Sequoia Capital, Capital Group, Coatue, GIC, ICONIQ, Amazon, Samsung, SK Hynix, Micron

2. Prometheus

Total Funding: $18B+ | Latest Valuation: $41B | Round: $12B Series B (June 11, 2026)

Prometheus is Jeff Bezos’s industrial AI startup, and its $12 billion Series B was one of the largest private rounds of the year. It valued the company at roughly $41 billion and pushed total funding past $18 billion in under a year of existence. Bezos co-leads the company as CEO alongside Vik Bajaj, a Verily co-founder, and this is his first operating role since leaving Amazon in 2021. Investors include JPMorgan, Goldman Sachs, BlackRock, DST Global, Arch Venture Partners, and Bezos himself.

The company is building what it calls an “artificial general engineer,” an AI system meant to take a complex physical product like a jet engine from first design all the way through to manufacturing. This is not a chatbot. It is aimed at the physical world, treating the whole design-to-build pipeline as one AI problem.

Prometheus has around 150 employees across San Francisco, London, and Zurich, and has hired from OpenAI, Google DeepMind, and Nvidia. It has not disclosed revenue and has named only Blue Origin as an early customer. Bezos told CNBC the round was raised largely to buy more compute.

Key Investors: Jeff Bezos, JPMorgan, Goldman Sachs, BlackRock, DST Global, Arch Venture Partners

3. DeepSeek

Total Funding: ~$7.4B | Latest Valuation: $50B+ | Round: $7.4B (June 2026)

DeepSeek raised more than 50 billion yuan, about $7.4 billion, in its first ever external funding round, at a valuation above $50 billion. For a company that ran on its founder’s own hedge fund money until now, that is a major shift. The round was reportedly led by Tencent, with around $1.5 billion, and battery giant CATL, with roughly $735 million. China’s state-backed AI fund was the only investor granted direct equity and voting rights.

The deal came with unusual terms. Outside investors put their money into a limited partnership run by founder Liang Wenfeng rather than into the company directly, took no voting rights, and agreed to a five-year lock-up. It is a structure built to keep control firmly in the founder’s hands.

DeepSeek stunned the industry in 2025 with low-cost models that matched far more expensive US systems. Its open-weight strategy keeps pressure on the whole market, forcing rivals to compete on cost, not just raw capability. It is doing all of this at a fraction of Anthropic’s or OpenAI’s valuation.

Key Investors: Tencent, CATL, China’s National AI Industry Investment Fund, Liang Wenfeng

4. Anduril Industries

Total Funding: $11B+ | Latest Valuation: $61B | Round: $5B Series H (May 13, 2026)

Anduril raised a $5 billion Series H on May 13, doubling its valuation to $61 billion in less than a year. The round was led by returning backers Thrive Capital and Andreessen Horowitz, and it came in about $1 billion higher than the raise the market expected. Total funding now sits above $11 billion across eight rounds.

The defense tech company builds AI-driven autonomous systems for the US military and its allies, from drones and counter-drone interceptors to its Lattice command software. CEO Brian Schimpf tied the raise to rising demand as the US and its allies deal with great-power competition and the shift toward autonomous warfare. The money is going into manufacturing, including its Arsenal-1 weapons factory in Ohio, plus R&D and infrastructure.

Anduril reported $2.2 billion in revenue for 2025, more than double the year before, and holds a $20 billion enterprise framework contract with the US Army. Its valuation has roughly doubled every year since 2022. The company still says an IPO is the goal, but not until the whole business is closer to profitable.

Key Investors: Thrive Capital, Andreessen Horowitz, Founders Fund, 8VC

5. DayOne Data Centers

Total Funding: ~$6.4B+ | Latest Valuation: $20B | Round: $4.5B Series C (June 5, 2026)

DayOne closed its Series C at $4.5 billion on June 5, more than doubling the round from its January first close, at a reported $20 billion valuation. The round was led by existing investors Coatue and Hillhouse, now the company’s two largest shareholders, with the Indonesia Investment Authority and Achi Capital Partners joining as new backers.

DayOne is a Singapore-based operator that builds and runs AI-ready data center campuses across Asia Pacific and Europe. It spun out of China’s GDS Holdings and has secured more than 1.5 gigawatts of capacity bookings since 2022. The new money funds expansion across Singapore, Malaysia, Indonesia, Thailand, Japan, and into Finland and Spain.

The round is a clear sign that AI funding has moved past software and into the physical layer that makes AI run. DayOne is also lining up a possible dual listing in Singapore and New York and chasing a corporate loan reported to be as large as $7 billion.

Key Investors: Coatue, Hillhouse, Indonesia Investment Authority, Achi Capital Partners

6. Isomorphic Labs

Total Funding: ~$2.7B | Latest Valuation: Undisclosed | Round: $2.1B Series B (May 12, 2026)

Isomorphic Labs raised $2.1 billion in a Series B on May 12, described by analysts as the second-largest biotech round ever. The round was led by Thrive Capital, with existing backers Alphabet and GV joined by new investors MGX, Temasek, CapitalG, and the UK government’s AI fund.

The London company was founded by Alphabet in 2021 to use AI to speed up drug discovery. It builds on the same science as AlphaFold 3, the Google DeepMind model that predicts how molecules fold and interact. Its in-house engine, IsoDDE, is meant to design potential drugs across many disease areas at once rather than chasing a single target.

The money will push IsoDDE forward and move Isomorphic’s own drug candidates toward the clinic. The size of the round, given the caution elsewhere in biotech funding, shows how much confidence sits behind AI-native drug design when it carries the DeepMind name.

Key Investors: Thrive Capital, Alphabet, GV, MGX, Temasek, CapitalG, UK Sovereign AI Fund

7. Moonshot AI

Total Funding: ~$3.8B | Latest Valuation: $20B | Round: $2B Series D (May 7, 2026)

Moonshot AI raised $2 billion in a Series D on May 7 at a $20 billion valuation, nearly doubling its worth from a few months earlier. The Beijing company is best known for its open-weight Kimi models, which have become some of the most used LLMs in the world for coding. Backers across its recent rounds include Alibaba, Tencent, Meituan, IDG Capital, and HSG.

The company was founded in 2023 by Yang Zhilin, a former Meta AI and Google Brain researcher. Its Kimi K2 series pushed China’s open coding models forward and posted benchmark numbers close to top US models at the time. By mid-2026 it had launched Kimi K3, one of the largest open AI models yet released.

Moonshot plans to use the new capital to cover the rising cost of training, buy compute, and grow its user base. The company is also reported to be preparing a Hong Kong IPO and removing its offshore structure, a move that would tie it more closely to China’s domestic markets.

Key Investors: Alibaba, Tencent, Meituan, IDG Capital, HSG, CPE

8. Baseten

Total Funding: ~$2B+ | Latest Valuation: $13B | Round: $1.5B Series F (Q2 2026)

Baseten raised $1.5 billion in a Series F, closing across two tranches at $13 billion and $11 billion valuations. The round was led by Altimeter Capital, Conviction, and Spark Capital, with Sands Capital and Wellington Management as co-leads, plus IVP, Greylock, Battery Ventures, and D.E. Shaw Ventures joining in.

The San Francisco company sits in the inference layer of AI. Its platform handles the messy work of running models in production: GPU orchestration, autoscaling, observability, billing, and developer tools. That lets companies mix frontier models with their own custom, fine-tuned ones without building the plumbing themselves.

Baseten’s raise was part of a clear Q2 trend. Capital rotated hard from model training toward inference serving, the part of the stack that actually delivers AI to end users at scale. Fireworks AI raised its own $1.5 billion round within weeks, making inference infrastructure one of the hottest categories of the quarter.

Key Investors: Altimeter Capital, Conviction, Spark Capital, Sands Capital, Wellington Management, IVP, Greylock, D.E. Shaw Ventures

9. NEURA Robotics

Total Funding: ~$1.68B | Latest Valuation: $7B | Round: $1.4B Series C (June 10, 2026)

NEURA Robotics announced a Series C of up to $1.4 billion on June 10 at a $7 billion valuation, which it calls the largest round ever for a full-stack robotics company. The German company, based in Metzingen near Stuttgart, builds what it calls cognitive robots, machines that combine hardware with AI so they can sense, learn, and adapt rather than just repeat fixed motions. Part of the funding is tied to performance milestones. The valuation is up from around $4 billion just eight months earlier.

The investor list says more than the dollar figure. The round was led by stablecoin issuer Tether, with Qualcomm, Amazon, Nvidia, Bosch, Schaeffler, the European Investment Bank, and imec.xpand all joining in. When Amazon and Nvidia show up in the same robotics round, it signals real belief in physical AI.

NEURA is best known for its 4NE-1 humanoid and its plan to bring cognitive robots into factories, warehouses, and eventually homes. It says its order book already tops $1 billion and is targeting 6,000 units in 2026, scaling toward millions by 2030. The raise makes it Europe’s most-funded humanoid maker in a race so far led by US and Chinese names.

Key Investors: Tether, Qualcomm, Amazon, Nvidia, Bosch, Schaeffler, European Investment Bank, imec.xpand

10. ICEYE

Total Funding: ~€663M prior + Series F | Latest Valuation: €10B+ (~$12B) | Round: €450M primary Series F, over €1B total (June 9, 2026)

ICEYE’s headline “€1 billion round” needs one clarification. The company raised €450 million ($520 million) in fresh primary Series F capital, led by General Atlantic. A secondary share sale, where existing shareholders sold stock to new investors, brought the total transaction above €1 billion. So the new money into the company is €450 million, not the full billion. The round valued ICEYE at more than €10 billion, about $12 billion, more than four times its December 2025 mark. Other investors include Nokia, the Qatar Investment Authority, TCV, and Finnish state-linked funds.

The Finnish company runs the world’s largest constellation of synthetic aperture radar satellites, which image the ground day or night and see through cloud cover. That makes it a key supplier of near-real-time intelligence for defense, disaster response, and insurance. It crossed €250 million in revenue in 2025 and was already profitable heading into the round.

Seven European governments have now bought sovereign satellite systems from ICEYE, and it delivered Poland’s constellation within 12 months of signing. Demand has surged since 2022 as European nations push for their own eyes in orbit. The company is doubling production from 50 satellites a year toward 100 by 2028.

Key Investors: General Atlantic, Nokia, Qatar Investment Authority, TCV, Solidium, Tesi, Varma, Ilmarinen, Lifeline Ventures

Q2 2026 by the Numbers

What This Quarter Actually Means

The record totals are only half the picture. A few shifts stand out once you look at where the money went.

One company still dominates, but it is a different one. Anthropic’s $65 billion round took close to a third of all global venture funding for the quarter and made it the most valuable private company on earth. In Q1 that spot belonged to OpenAI. The frontier model race now has a new leader, and the gap between the top labs and everyone else keeps widening.

The money spread out. Q1 was almost entirely about foundation model labs. Q2 pushed capital into the whole ecosystem around AI: defense with Anduril, robotics with NEURA, drug discovery with Isomorphic, data centers with DayOne, and space intelligence with ICEYE. Investors have moved from backing the biggest models to funding everything those models need and everything they can be pointed at.

Inference became the new battleground. Baseten and Fireworks AI both raised $1.5 billion within weeks of each other. As companies move from testing AI to actually running it in production, the value is shifting from training models to serving them fast and cheap. That is where a lot of the next fight will happen.

China is competing on its own terms. DeepSeek and Moonshot AI together raised close to $10 billion, both with open-weight strategies and deal structures built to keep founders in control. They are doing it at a fraction of the valuations their US rivals command, and their open models keep cost pressure on the entire market.

The exit window opened, then got tighter. Q2 was the biggest quarter ever for billion-dollar acquisitions, headlined by SpaceX buying Cursor maker Anysphere for $60 billion, the largest purchase of a venture-backed startup on record. But SpaceX’s own IPO slipped below its offer price within weeks, OpenAI pushed its listing to 2027, and Databricks pulled out of the 2026 queue. The private money is still unlimited. The public market suddenly looks narrower than it did in early June.

The post Top 10 Biggest AI Funding Rounds of Q2 2026 appeared first on AI Funding Tracker.

This guide covers exactly where spreadsheets fail, why diligence is where the cracks show, and how to tell you have already outgrown yours.

Why a spreadsheet works at first

At the start there is not much to track. A couple of founders, maybe an advisor, a clean split. A spreadsheet handles that easily, and reaching for dedicated software on day one would be overkill.

So this is not an argument against ever using a spreadsheet. It is an argument about timing. The spreadsheet is fine right up until the number of moving parts grows past what a person can track by hand without slipping. That point arrives sooner than most founders expect.

Where spreadsheets actually break

The failures are not dramatic. They are small errors that sit unnoticed until something forces a close look. Here is what tends to go wrong.

Version control. You email the spreadsheet to your lawyer. They edit it. You edit your own copy at the same time. A co-founder saves a third version. Now there are three files and no clear answer to which one is real. This sounds minor until an investor asks for the current cap table and you are not sure which file to send.

Formula errors that compound. One wrong cell reference early on flows into every calculation after it. Because the spreadsheet still produces a confident looking number, nobody notices. The error only surfaces when someone checks the math against the actual signed documents.

SAFE and note conversion. This is the big one. When you have raised on several SAFEs at different caps, working out how they convert at a priced round is genuinely hard to do by hand. Get it wrong and your ownership numbers are off, your investors’ numbers are off, and the mistake is baked into the round before anyone catches it.

Option grants and vesting. Every grant has a start date, a cliff, a vesting schedule, and sometimes an early exercise or an early departure. Tracking all of that across a growing team in a spreadsheet means updating it by hand every single time, and one missed update means your fully diluted math is quietly wrong.

No audit trail. A spreadsheet does not remember who changed what, or when. When an investor asks you to prove a number is correct, you cannot point to a record. You can only point to more spreadsheet.

Why Series A is where it snaps

Seed rounds are often light on diligence. A SAFE or a priced seed can close without anyone stress testing your cap table.

Series A is a different animal. Now there is a lead investor writing a large check, and their lawyers go through your ownership records in detail. They check that every share, option, SAFE, and note is accounted for. They check that the math is internally consistent. They check that what the cap table says matches what the signed documents say.

If your spreadsheet has drifted from reality, this is where it shows. And it shows at the exact moment you have the least room to deal with it, in the middle of a live round with a term sheet on the clock.

What a broken cap table actually costs

The cost is rarely just the error itself. It is what the error does to the round.

It slows you down. Diligence stalls while everyone works out what the real numbers are, and a deal that felt done starts to drag.

It can change the terms. If the cap table was wrong in a way that affected ownership, fixing it can mean renegotiating who owns what, which is an awkward conversation to have with money on the table.

It costs legal fees. Untangling a messy cap table under deadline pressure is exactly the kind of work lawyers bill a lot of hours for.

And it costs trust. A sloppy cap table makes an investor wonder what else is sloppy. That is not the impression you want to give the person about to fund your company.

Signs you have already outgrown the spreadsheet

You do not need to wait for a round to know it is time. Any one of these is a good signal.

You have granted options to employees. You have raised on more than one or two SAFEs. You are not fully sure your spreadsheet matches your signed documents. You have more than one version of the file floating around. You are planning to raise in the next 6 to 12 months. Or you had to stop and think for more than a few seconds about who owns what, fully diluted, right now.

If any of those are true, the spreadsheet has already turned into a liability. The only question left is whether you fix it calmly now or under pressure later.

What software actually fixes

Dedicated cap table software is not magic, but it removes the specific failure points above.

It keeps one source of truth, so there is never a question of which file is real. It does the dilution and conversion math for you, including the hard SAFE conversions, so the numbers stay consistent by default. It tracks every option grant and vesting schedule automatically. It keeps a record of changes, so you can show an investor exactly how a number came to be. And when diligence starts, it produces a clean export in minutes instead of a weekend of scrambling.

The result is that your cap table stops being something you hope is right and becomes something you know is right. During a raise, that difference is worth a lot.

The bottom line

Spreadsheets are the right tool for the first few months and the wrong tool the moment real equity starts moving. The break rarely announces itself. It waits until diligence, when the cost of a mistake is highest, and then it shows up all at once.

The move is simple, and it feels optional right up until it isn’t. Switch to real software once you have your first option grants or your first outside money, well before your Series A, so that when the lawyers come looking, your cap table is boring, clean, and correct.

The post Why Spreadsheet Cap Tables Break at Series A appeared first on AI Funding Tracker.

This guide walks through the parts of startup equity that actually matter, in plain terms, so you can make decisions now that your future investors and your future self will thank you for.

What a cap table is

A cap table, short for capitalization table, is the list of every ownership stake in your company. That includes founder shares, employee options, investor shares, SAFEs, convertible notes, and anything else that can turn into equity later.

In the first few months a spreadsheet does the job. By the time you have a few employees with options, a couple of SAFEs at different caps, and one priced round behind you, the spreadsheet becomes a risk. One wrong formula or one outdated version can throw off your ownership math, and that error tends to surface at the worst possible time, during investor diligence.

The building blocks of your equity

Founders often blur these together. Keeping them separate is most of the job.

Common stock. This is what founders and employees hold. It is plain ownership with no special rights attached.

Preferred stock. This is what investors get in a priced round. It comes with extra rights, usually around what happens to their money in a sale or a shutdown.

Options. These give an employee the right to buy shares later at a fixed price, called the strike price. Options are the standard way to give equity to your team.

Option pool. This is a block of shares you set aside for future hires. It usually sits somewhere between 10 and 20 percent of the company, and investors often ask you to top it up right before they invest. That top up comes out of your ownership, not theirs, so it is worth negotiating.

SAFEs and convertible notes. These are ways to raise money quickly without setting a price for your company yet. They convert into shares later, usually at your next priced round. More on how they hit your cap table below.

How ownership gets divided

At the start, founders own everything, split however you agree. From there, every round of hiring and fundraising chips away at that number. This is called dilution, and it is normal. Owning a smaller slice of a bigger, better funded company is the whole point.

Here is the rough shape of it. Founders start at 100 percent. You carve out an option pool for the team. Then each raise hands a chunk to investors. A seed round often costs founders somewhere in the range of 10 to 20 percent, and a Series A commonly lands in a similar or slightly higher range, though the real number depends entirely on how much you raise and at what valuation. Treat those as rough guides, not promises.

The mistake to avoid is thinking about any single round on its own. Dilution stacks across rounds. A generous option pool plus a big seed plus a big Series A can add up faster than founders expect.

Vesting and the 83(b) election

Vesting means you earn your shares over time instead of getting them all at once. The standard schedule is four years with a one year cliff. The cliff means you earn nothing for the first year, then 25 percent at the one year mark, then the rest month by month after that. This applies to founders too, not just employees, and investors expect to see it.

Here is the part that trips people up. If you receive stock that is subject to vesting, you can file something called an 83(b) election with the IRS. It tells the IRS to tax you on the stock now, while it is worth almost nothing, instead of later as it vests and grows in value. For most founders this is a large tax saving.

The catch is the deadline. You have 30 calendar days from the date the stock is granted to file, and there are no extensions, ever. Miss it and you cannot fix it. As of 2025 you can file online through the IRS using Form 15620, which is simpler than the old paper method, but the 30 day clock is exactly the same. File within the first week or two to give yourself margin.

SAFEs and convertible notes

Early on, most US startups raise on SAFEs. A SAFE lets an investor give you money now in exchange for shares later, without either side having to agree on a company valuation today. It converts into equity at your next priced round.

The detail that matters most is the valuation cap. The cap sets the highest company value at which the investor’s money converts into shares. A lower cap means the investor gets more shares when conversion happens. Since 2018 the standard version from Y Combinator is the post-money SAFE, which calculates the investor’s ownership as a percentage of your company after all SAFEs convert. Post-money SAFEs are easier to add up, but they also dilute founders more than the older pre-money version, and that extra dilution is easy to underestimate when you stack several of them.

Convertible notes work in a similar way but are structured as debt, with interest and a maturity date. They were more common before SAFEs took over. If you are choosing today, most US seed investors expect a SAFE.

The thing to watch with both is what happens at conversion. Several SAFEs at different caps converting into the same round can move your ownership more than you planned. Model it before you sign, not after.

Priced rounds and dilution

A priced round is when you and your investors agree on an actual value for the company and they buy preferred stock at that price. This is usually your Series A and beyond, though some seed rounds are priced too.

Two numbers matter here. The pre-money valuation is what your company is worth before the new money goes in. The post-money valuation is the pre-money plus the new investment. The investor’s ownership is their check divided by the post-money valuation. So a 3 million dollar investment at a 12 million dollar post-money valuation buys 25 percent of the company.

This is also the round where all those earlier SAFEs convert into real shares. It is common for founders to be surprised at how much of the company is already spoken for once the SAFEs land and the new option pool is added on top.

409A valuations

A 409A valuation is an independent appraisal of what your common stock is worth. You need one to set the strike price on employee options at fair market value, which keeps those options from creating a tax problem for your team.

You need a 409A before you grant options after a priced round, and you need a fresh one every 12 months, or sooner if something material happens such as a new funding round. Some cap table platforms include 409A valuations in their pricing. Others send you to a third party that charges a few thousand dollars each time. If you plan to grant options in your first year, that difference adds up quickly.

QSBS and a tax rule worth knowing early

QSBS, short for Qualified Small Business Stock, is one of the biggest tax breaks available to founders and early employees. If your company qualifies and you hold your shares long enough, you can exclude a large share of the gain from federal tax when you sell. It applies to stock in US C-corporations, and some business types are excluded, so it is not automatic.

This rule changed in July 2025 under the One Big Beautiful Bill Act, and the new version is more generous, but only for stock issued after July 4, 2025. Here is the shape of it for that newer stock. Hold for at least three years and you can exclude 50 percent of the gain. Hold four years and it is 75 percent. Hold five years or more and it is 100 percent. The cap on the excluded gain rose to 15 million dollars, up from 10 million, and the size limit on a qualifying company rose to 75 million dollars in gross assets, up from 50 million.

Stock issued on or before July 4, 2025 still follows the old rules, which means a five year hold for any exclusion and a 10 million dollar cap. If you hold shares from both before and after that date, you now have two separate tax situations to track, which is one more reason to keep clean records. This is real money at exit, so it is worth a conversation with a tax advisor early rather than late.

How founders break their own cap tables

The same handful of mistakes show up again and again.

Missing the 83(b) deadline. The 30 day window is unforgiving and the tax cost is real.

Handshake equity. Promising someone shares in a conversation, or over email, with no paperwork behind it. When it comes time to formalize the grant, the two sides rarely remember the terms the same way.

Stacking SAFEs without modeling them. Raising on several SAFEs at different caps and only discovering the combined dilution at the priced round.

Letting the cap table go stale. Not recording option grants, exercises, or transfers as they happen, then trying to rebuild the record under deadline pressure during diligence.

None of these are hard to avoid. They are just easy to postpone.

When to move off a spreadsheet

A spreadsheet is fine at the very start. The point where it stops being fine is roughly your first real option grants or your first outside investor, because that is when the number of moving parts and the cost of an error both jump at once.

Cap table software automates the dilution math, tracks vesting, keeps your 409A and filings in one place, and gives investors a clean export during diligence. The market is competitive in 2026, with strong options at every stage and price point. One recent shift is worth noting if you are picking a tool now. In August 2025, AngelList stopped taking new customers for its standalone Stack cap table product and named Pulley and J.P. Morgan Workplace Solutions as migration partners, so Stack is no longer a starting option for new companies.

If you want the full breakdown of which tool fits which stage, read our guide to the best cap table management software.

The bottom line

Equity is not complicated once you separate the parts. Shares, options, pools, SAFEs, and priced rounds each do one job. Vesting and the 83(b) election protect you early. The 409A keeps your option grants clean. QSBS can save you a lot at the end if you set it up right. And keeping the record accurate, from day one, is what turns your next raise into a formality instead of a fire drill.

Start simple, keep it current, and move to real software the moment your equity runs longer than a single page.

The post AI Founder’s Guide to Equity and Cap Tables appeared first on AI Funding Tracker.

The good news is that the cap table software market in 2026 is genuinely competitive, with strong options at every stage and price point. The category is expected to be worth $6.3 billion this year and is projected to reach $15 billion by 2033, driven primarily by startups increasing equity compensation as a cost management tool in a tighter funding environment.

This guide covers the six platforms that consistently appear on shortlists across pre-seed through pre-IPO, with verified pricing, honest trade-offs, and a clear recommendation on which stage each tool fits best.

What Cap Table Software Actually Does

A cap table tracks every security a company has issued- common stock, preferred stock, options, warrants, convertible notes, and SAFEs alongside the people and entities that hold them. The software automates dilution math, vesting schedules, 409A valuations, and the paperwork investors and acquirers require during diligence.

Early on, a spreadsheet works. By Series A, with multiple security types, vesting cliffs, options grants, and SAFEs converting at different valuation caps, a spreadsheet becomes a liability. A single formula error or version conflict in a shared sheet can cascade into investor renegotiations or a delayed close.

What to Look For Before You Choose

Stage fit. A pre-seed tool that is free but lacks scenario modeling is wrong for a Series B company. An enterprise platform with quote-based pricing is wrong for a two-person team pre-raise.

409A support. You need a 409A valuation before issuing any new options after a priced round, and every 12 months thereafter. Some platforms include it natively. Others refer you to third-party providers at $2,000 to $4,000 per valuation. That difference compounds quickly.

Investor familiarity. Some platforms are deeply embedded in the diligence workflows of US law firms and institutional investors. If your investors expect a Carta-exported cap table, running on a platform they have never seen creates friction at the worst possible moment.

Geography. US-centric platforms handle 409A, ASC 718, QSBS, and IRS filings well but have limited support for European equity structures: EMI schemes, CSOP options, VSOP in Germany, BSA-AIR in France, and GDPR requirements for employee data. European founders should not force a US-first tool into their compliance stack.

Migration support. You will switch platforms at least once. Free, clean migration support is not a bonus feature; it is a practical requirement.

The 6 Best Cap Table Management Platforms in 2026

1. Eqvista

Best for: Pre-seed through late-stage companies, including Series funding to Pre-IPO that need both cap table management and bundled 409A valuation at the lowest total cost

• Pricing: Free up to 20 shareholders. Premium plan at $2 per shareholder per month. 409A Valuation plan starts at $990 per year with unlimited valuation updates.
• G2 Rating: 4.9 out of 5 (148 reviews), ranked #1 in G2’s Equity Management Software category
• HQ: United States

Eqvista is the strongest value proposition in the market for startups at any stage that want cap table management and 409A under one roof without the pricing structure of Carta. The free tier covers up to 20 shareholders with real-time cap table tracking, SAFE and convertible note management, and basic vesting schedules. The Premium plan, at $2 per shareholder per month, is the most transparent pricing model in this category. There are no hidden add-on fees for features that other platforms charge separately.

The platform’s distinguishing feature is its Real-Time Company Valuation technology, which sits alongside traditional 409A valuations conducted by in-house NACVA-certified professionals. Every 409A includes lifetime audit support and full IRS defense. The plan also covers ASC 718 reporting, 83(b) elections, and QSBS attestation, all integrated with the cap table at no additional cost. G2 reviewers from healthcare, technology, and consulting consistently describe savings of $8,000 to $12,000 per year compared to traditional 409A firms or Carta’s equivalent tiers.

Scenario modeling and waterfall analysis are frequently cited by G2 users as standout capabilities. The platform supports round modeling, dilution simulations, and funding scenario planning, features that help founders understand their ownership structure before entering investor conversations.

The investor CRM is less specialized than some competitors, and the platform’s investor brand recognition in institutional diligence workflows does not yet match Carta. For companies whose investors specifically request Carta-format exports, that is worth factoring in.

Verdict: The best starting point for pre-seed to later-stage funded companies, and a credible alternative to Carta for growth-stage teams that want to reduce their annual equity management spend significantly.

2. Carta

Best for: Series A and beyond; companies that need the deepest institutional trust, the broadest feature set, and fund administration under one roof

• Pricing: Launch plan free for up to 25 stakeholders. Build plan approximately $2,800 per year. Grow and Scale plans step up from there. Enterprise is quote-based.
• G2 Rating: 4.4 out of 5 (224 reviews)
• HQ: New York, NY
• Company scale: $7.4B valuation, $500M ARR (2025), 40,000+ companies on platform

Carta is the category-defining platform for US private company equity. It serves 40,000-plus companies and runs the 409A engine that institutional VCs and acquirers encounter most frequently during diligence. At Series A, when your investors are running a structured diligence process and your law firm is managing cap table documentation, being on Carta removes friction that other platforms can create, simply because Carta is what the ecosystem expects.

The feature set is the broadest in the market: cap table management, 409A valuations, equity plan administration, board consent workflows, fund administration for VC funds, LP portals, and a secondary transaction marketplace. No other single vendor covers this much surface area. For companies planning a liquidity event, an institutional round, or fund administration alongside cap table management, Carta’s consolidated approach avoids the overhead of managing multiple vendors.

The trade-off is cost. At the Build tier, 409A valuations cost $2,000 to $4,000 as add-ons beyond the base plan. G2 reviews and buyer comparisons consistently note that Carta’s pricing escalates steeply with stakeholder count and feature depth, with some mid-stage companies reporting annual costs between $14,000 and $20,000. Onboarding and support reviews are more mixed than the platform’s functionality warrants.

Verdict: The default choice for Series A and beyond, especially if your institutional investors, law firm, or board have existing Carta workflows. Overkill and overpriced for pre-seed.

3. Pulley

Best for: YC-backed and early-stage US startups that want a clean, founder-friendly experience with transparent pricing

• Pricing: Startup plan at $1,200 per year for up to 25 stakeholders. Growth plan at $3,600 per year for up to 40 stakeholders, including 409A. Higher tiers are quote-based.
• G2 Rating: 4.6 out of 5
• HQ: San Francisco, CA

Pulley built its reputation as the founder-friendly Carta alternative, and it earns that reputation through transparent pricing, fast onboarding, and a clean user interface that does not require implementation support to operate. The Startup plan at $1,200 per year is meaningfully cheaper than comparable Carta tiers, and the Growth plan includes 409A valuation support at a price point most seed-stage companies can absorb.

The platform handles the standard US equity structure well: SAFEs, convertible notes, common and preferred stock, standard vesting schedules, and scenario modeling for fundraising rounds. AngelList named Pulley as an official migration partner when Stack sunset in August 2026, which accelerated Pulley’s market position among pre-seed and seed stage teams who were previously on AngelList.

Pulley’s limitations are at the edges. Complex modeling scenarios, IFRS reporting, and international equity structures are not where the platform is strongest. Very early pre-seed teams paying $1,200 per year for features they will not use for twelve months may find Eqvista’s free tier a better starting point.

Verdict: The cleanest Carta alternative for US seed companies, particularly for founders who want transparent pricing and fast setup without enterprise complexity.

4. Ledgy

Best for: European startups, UK-incorporated companies, and any team with cross-border shareholders requiring multi-currency cap table management and EU compliance

• Pricing: Quote-based, typically 10 to 20% less than Carta for comparable European stakeholder structures
• G2 Rating: 4.7 out of 5
• HQ: London, UK

Ledgy is the clear category leader for European equity management. US-centric platforms like Carta and Pulley have limited support for EMI schemes, CSOP options, VSOP in Germany, BSA-AIR in France, and GDPR requirements for employee equity data. Ledgy was built specifically for these structures and handles multi-currency cap tables, automated EU equity plan management, IFRS 2 reporting, and localized compliance across European regulatory environments.

The platform supports investor reporting, scenario modeling, and vesting management at a standard that matches Pulley and Carta for European use cases. Its investor and employee portals are highly regarded in G2 reviews for transparency and usability. Ledgy does not natively offer 409A valuations, which is expected given its European focus, but for UK and EU companies that do not require US-format valuations, this is not a gap.

The limitation is straightforward: Ledgy does not compete on US-specific features. For US-only companies, the name recognition with US institutional investors and law firms does not match Carta or Pulley. 409A valuations are not natively offered.

Verdict: The default choice for European and UK-incorporated startups. Non-negotiable if your equity structures include EMI, CSOP, VSOP, or BSA-AIR.

5. Cake Equity

Best for: APAC and Australia-headquartered startups, and early-stage global teams that want lightweight equity management with strong onboarding

• Pricing: Startup-friendly tiers starting around $80 per month. Quote-based at growth stage.
• G2 Rating: 4.8 out of 5
• HQ: Sydney, Australia

Cake Equity is the strongest platform in the APAC market and has built a reputation for customer support that consistently outperforms larger competitors in G2 reviews. The platform covers cap table management, ESOP administration, and investor management, with an education-first approach that makes it particularly useful for founding teams who are new to equity management. G2 ratings of 4.8 cite fast onboarding, responsive support, and clean workflows as the primary drivers.

Startups switching from Carta to Cake Equity have reported saving $3,000 to $16,000 per year depending on stakeholder count and feature usage. The platform is available globally but is most deeply integrated with APAC legal and compliance frameworks.

The trade-off is feature depth at later stages. Carta and Pulley have stronger US-specific functionality for companies planning an institutional round with US investors, and the platform’s name recognition in US diligence workflows is limited.

Verdict: The best choice for APAC-headquartered startups and a strong Carta alternative for cost-conscious early-stage teams globally.

6. Shareworks by Morgan Stanley

Best for: Late-stage and pre-IPO companies that need institutional-grade equity plan administration and integration with wealth management services

• Pricing: Quote-based, enterprise pricing
• HQ: United States

Shareworks targets the high end of the market: late-stage private companies, pre-IPO teams, and public companies that need equity plan administration at institutional scale. The Morgan Stanley parentage means deep integration with wealth management, secondary transactions, and liquidity programs for employees and investors. For a company with hundreds of employees holding options and institutional investors managing position sizes, Shareworks provides a level of operational infrastructure that consumer-facing platforms are not built to deliver.

The trade-off is obvious: Shareworks is enterprise software priced and built for enterprise complexity. It is not relevant for pre-seed, seed, or early Series A companies, and the implementation timeline is not compatible with the speed that most early-stage teams need.

Verdict: Relevant only at late-stage and pre-IPO. If you are reading this guide because you are evaluating platforms for a seed or Series A company, Shareworks is not your platform.

Head-to-Head: Pricing Comparison

One Thing That Changed in 2026: AngelList Stack Is Gone

AngelList stopped accepting new Stack cap table customers in August 2026. Existing customers can remain on current plans, but the product is in maintenance mode. AngelList named Pulley and J.P. Morgan Workplace Solutions as official migration partners. JPMWS is offering free service for up to 100 stakeholders for three years to migrating teams. If you are currently on Stack and planning a round in 2026 or 2027, migrate before you enter diligence. Moving a cap table mid-round is painful.

How to Choose Based on Stage

Pre-seed (under $1M raised): Start with Eqvista’s free tier or Carta Launch. Both cover the basics at zero cost. Eqvista’s free plan extends to 20 shareholders; Carta Launch covers 25. Eqvista is the better choice if you anticipate needing 409A support in the next 12 months.

Seed and early Series A (US): Pulley at $1,200 per year or Eqvista’s Premium plan. Pulley has stronger name recognition in YC-adjacent circles. Eqvista wins on total cost if 409A is a near-term requirement.

Series A and beyond (US): Carta is the default, primarily because of ecosystem integration with investors and law firms. Eqvista and Carta at 20 to 40% lower cost is a credible alternative if your investors do not specifically require Carta-format outputs.

European or UK-incorporated: Ledgy, without exception. Do not try to manage EMI or CSOP options in a US-first platform.

The Bottom Line

Cap table software is not a vanity purchase. The cost of a spreadsheet error at Series A diligence, recalculating fully diluted ownership with three simultaneous SAFE conversions and a previous option pool expansion, is typically several legal hours and a delayed close. The annual cost of the right platform is a rounding error by comparison.

Start with the free tier of whichever platform fits your stage. Migrate to a paid plan when you hit your first option grant or your first outside investor. Do not wait until your Series A data room request arrives to realize your cap table is on a spreadsheet that has not been reconciled in six months.

The platform itself matters less than the discipline of keeping it updated. But the right platform makes that discipline considerably easier to maintain.

The post 6 Best Cap Table Management Software in 2026 (Based on Reviews) appeared first on AI Funding Tracker.

The post AI Startup Funding News Today appeared first on AI Funding Tracker.

That’s the story of Nvidia’s investing over the past nine months: bigger checks, faster pace, and then a deliberate step back from picking sides in an AI lab rivalry that’s turned unusually public.

Nvidia Investment Activity by the Numbers

• 67 venture deals in full-year 2025 (excluding NVentures), up from 54 in 2024, per PitchBook
• NVentures alone made 30 deals in 2025 and roughly 20 more in the first five months of 2026
• Biggest single commitments: OpenAI ($30 billion final, down from an earlier $100 billion framework), Anthropic ($10 billion), xAI (up to $2 billion equity)
• Corporate development-led AI equity investments alone exceeded $40 billion in just the first four months of 2026
• Market cap: roughly $4.7-4.8 trillion as of early July 2026, briefly touching $5 trillion+ in late May, making Nvidia the world’s most valuable public company
• Investment focus remains AI model companies, cloud infrastructure, and emerging tech (quantum computing, robotics, biotech)

Nvidia still frames its investing around the same three buckets: AI model companies, cloud infrastructure providers, and frontier technology like quantum computing, fusion energy, and robotics. What’s changed is the appetite for writing nine and ten figure checks into the two companies that arguably matter most to its own future.

The Big Shift: Nvidia Steps Back From OpenAI and Anthropic

This is the most important development since our last update, and it reframes everything else in this article.

What happened: In November 2025, Nvidia and Microsoft jointly announced they would invest up to $10 billion and $5 billion respectively in Anthropic, alongside Anthropic committing to purchase $30 billion of Azure compute capacity. In February 2026, Nvidia finalized a $30 billion investment in OpenAI as part of a roughly $110 billion funding round, a steep reduction from the up-to-$100 billion framework announced the previous September.

Then, on March 4, 2026, speaking at the Morgan Stanley Technology, Media and Telecom conference, Jensen Huang said both investments were likely to be Nvidia’s last major equity checks into either company. His stated reason: both OpenAI and Anthropic were expected to go public later in 2026, and the private-investment window closes once that happens.

The more interesting story sits underneath that explanation. Around the same time, tensions between OpenAI and Anthropic became unusually public. The US government reportedly ordered federal agencies to stop using Anthropic’s technology after the company declined to let Claude be deployed for autonomous weapons or mass domestic surveillance. Within hours, OpenAI announced its own Pentagon deal, which Anthropic publicly called “mendacious.” Nvidia found itself holding significant equity in both companies just as they were pulling in opposite directions, commercially and politically. Huang later dismissed the idea that bad blood drove the pullback, but the timing raised questions of its own.

Why it matters for the ecosystem: Nvidia effectively said it doesn’t want to keep picking financial sides between its two biggest customers. Analysts expect this to ripple outward, tighter Series B/C funding environments for AI model companies more broadly, as the signal that even Nvidia sees diminishing returns in AI lab equity spreads through venture capital. It also puts more scrutiny on the remaining web of cross-investments, Microsoft’s stake in OpenAI, Amazon’s stake in Anthropic, Google’s positions in multiple labs, all of which face growing regulatory attention.

Top Nvidia Startup Investments, 2025-2026

1. OpenAI: $30 Billion Final Investment (February 2026)

• Original framework (September 2025): up to $100 billion across 10 tranches
• Actual investment finalized: $30 billion, as part of a roughly $110 billion funding round
• Round valued OpenAI at $852 billion post-money by the time it closed in March 2026
• Structure still ties partly to infrastructure deployment and GPU lease arrangements rather than upfront purchases

The headline number from our last update, up to $100 billion, never fully materialized. Nvidia scaled its actual commitment down to $30 billion, and Huang has said publicly that a $100 billion outcome is “probably not in the cards.” OpenAI’s own valuation still climbed sharply regardless, closing its round at $852 billion with Amazon and SoftBank also anchoring.

2. Anthropic: $10 Billion Investment (November 2025)

• Nvidia commitment: up to $10 billion, alongside a separate $5 billion commitment from Microsoft
• Anthropic committed to purchase $30 billion of Microsoft Azure compute capacity, plus additional capacity up to 1 gigawatt
• Anthropic and Nvidia also announced a deep technology partnership to optimize Claude models for Nvidia architecture and vice versa
• Anthropic’s subsequent Series H in May 2026 raised $65 billion at a $965 billion post-money valuation, overtaking OpenAI’s private mark

This is the biggest new entry since our last update. Anthropic wasn’t previously a named Nvidia portfolio company; it now represents one of Nvidia’s two largest single-company bets, alongside OpenAI. Notably, both Nvidia and Microsoft’s investments arrived as Claude became the only frontier model available across all three major cloud platforms.

3. xAI: Up to $2 Billion Equity in a $20 Billion Round (October 2025)

• Total round size: $20 billion
• Nvidia equity investment: up to $2 billion
• Structure: $7.5 billion equity plus $12.5 billion debt, largely through a special purpose vehicle for GPU purchases
• Chips support xAI’s Colossus 2 data center buildout in Memphis

No major changes here since October, though xAI’s broader capital picture has grown more complex as it consolidated with parts of Musk’s other ventures ahead of its own planned IPO.

4. UK AI Startup Ecosystem: £2 Billion Commitment (September 2025)

Unchanged in structure. Nvidia’s £2 billion (about $2.6 billion) commitment continues flowing through partner VCs (Accel, Air Street Capital, Balderton, Hoxton Ventures, Phoenix Court) into London, Oxford, Cambridge, and Manchester. Nvidia has since extended similar VC Alliance partnerships to European firms including Accel, Elaia, Partech, and Sofinnova, offering DGX Cloud Lepton marketplace credits to portfolio companies across the continent.

5. Mistral AI: €1.7 Billion Series C (September 2025)

Unchanged since our last update. Nvidia remains an investor alongside lead backer ASML, with Mistral valued at €11.7 billion ($13.8 billion).

6. CoreWeave: Nvidia Doubles Down With a Second Investment (Early 2026)

• Original stake: 7% (24.2 million shares), worth roughly $2 billion at IPO
• New: an additional $2 billion investment in early 2026, on top of a prior $6.3 billion agreement to purchase CoreWeave’s unused computing capacity through 2032

Nvidia didn’t just hold its post-IPO CoreWeave stake, it added to it. The combined equity-plus-capacity-purchase relationship makes CoreWeave one of the clearest examples of Nvidia’s circular financing model in practice: invest in the company, then buy back its compute.

7. Synopsys: $2 Billion Investment (Late 2025)

A new addition to the portfolio since our last update. Nvidia invested $2 billion into chip-design software maker Synopsys by the end of 2025, deepening ties between Nvidia’s hardware roadmap and the design tools used to build it.

8. Nebius: $2 Billion Investment (March 2026)

Another new entry. Nvidia invested $2 billion in GPU cloud provider Nebius in March 2026, adding a third major cloud-capacity relationship alongside CoreWeave and Nscale.

9. Nscale: $433 Million SAFE (October 2025)

Unchanged. Nvidia’s £500 million commitment to the UK cloud provider continues, following Nscale’s earlier $1.1 billion raise.

10. Legora: $50 Million Series E (April 2026)

Nvidia’s first direct investment in legal AI. Legora, a Swedish legal-AI startup valued at $5.6 billion and built on Anthropic’s Claude, raised the round days after NVentures also backed Israeli clinical AI company Aidoc’s $150 million Series E, signaling a deliberate push into vertical, inference-heavy application companies rather than only infrastructure and foundation models.

Additional Notable Investments Since October 2025

• PhysicsX and Alice & Bob: NVentures backed physics-simulation platform PhysicsX and French quantum computing startup Alice & Bob in the first half of 2026, continuing Nvidia’s quantum and simulation hedge alongside existing bets like PsiQuantum and Quantinuum
• Generalist AI: $400 Million Round (June 2026): Nvidia and Fei-Fei Li backed this AI robotics startup’s round alongside Radical Ventures and Bezos Expeditions
• XBOW: $35 Million Series C Extension (May 2026): NVentures joined Accenture and DNX Ventures in backing this offensive security-testing startup
• Redwood Materials: Nvidia invested in the battery and energy-storage company to help power AI data centers, an increasingly common pattern as compute buildouts strain grid capacity
• Quantinuum: Went public on Nasdaq in June 2026, another exit for Nvidia’s quantum computing bets, following Cerebras’s IPO earlier in the year

Investment Categories and Distribution (Updated)

By Sector:

• AI Model Companies: still the largest share, though Anthropic’s addition and OpenAI’s reduced check size shift the dollar-weighted mix
• Cloud Infrastructure: growing, with new entries (Nebius, continued CoreWeave) alongside existing bets (Crusoe, Lambda, Nscale)
• Emerging Tech: expanding fastest in deal count, quantum computing (Alice & Bob, PsiQuantum, Quantinuum), fusion (Commonwealth Fusion), and energy storage (Redwood Materials)
• Vertical Applications: a newer category taking shape, with legal AI (Legora), clinical AI (Aidoc), and robotics (Generalist AI, Figure AI) all drawing fresh checks in 2026

By Geography: US deals remain the majority, but Nvidia’s European push, through both direct investment and the VC Alliance program, has visibly expanded since late 2025, particularly in France, the UK, and the broader EU AI ecosystem.

How Nvidia’s Investment Strategy Works

The Circular Financing Model, Under More Scrutiny

The core pattern hasn’t changed: Nvidia invests in a startup, the startup uses much of that capital to buy Nvidia GPUs or lease Nvidia-powered cloud capacity. CoreWeave remains the clearest example, an equity stake plus a $6.3 billion compute-purchase agreement running through 2032. Critics have long called this circular financing, and the scrutiny has intensified as the dollar figures involved have grown into the tens of billions per deal.

What’s different now is that Nvidia’s own leadership has started publicly distancing itself from at least part of this pattern, specifically the practice of holding equity in AI labs that increasingly compete with each other and, at times, with Nvidia’s own DGX Cloud and NIM inference offerings. Whether Nvidia extends that same caution to its cloud-infrastructure bets (CoreWeave, Nebius, Nscale, Lambda) remains to be seen; those relationships show no signs of slowing.

Two-Layer Investment Structure

Nvidia’s dealmaking effectively runs through two teams operating at very different scales:

1. Corporate Development (led by Vishal Bhagwati): handles the largest strategic checks, OpenAI, Anthropic, Synopsys, CoreWeave, Nebius, xAI. This is where the $30-billion-plus commitments live.
2. NVentures (led by Mohamed “Sid” Siddeek, a two-person team): handles earlier-stage, higher-volume deals across nearly every sector Nvidia can technically reach, from quantum computing to legal AI to robotics. NVentures has backed 80+ companies since 2021, including 20 that reached unicorn status.

Revenue and Market Position

Nvidia Financial Performance (Q1 FY2027, reported May 20, 2026):

• Revenue: $81.6 billion, up 85% year over year and 20% quarter over quarter, a record
• Data center revenue: $75.2 billion, up 92% year over year, also a record
• GAAP gross margin: 74.9%
• The board approved an additional $80 billion share buyback authorization in May 2026, on top of returning roughly $20 billion to shareholders in the quarter alone
• Quarterly dividend increased from $0.01 to $0.25 per share, effective late June 2026

Customer concentration remains a live risk factor. Nvidia’s top customers still represent a large share of data center revenue, and CEO commentary continues to acknowledge that concentrated demand from a handful of hyperscalers and AI labs, several of which Nvidia also holds equity in, cuts both ways: it drives extraordinary growth, but it also means Nvidia’s fortunes are tightly bound to a small number of counterparties’ spending decisions.

Comparison With Other Tech Giants

Microsoft: Beyond its long-standing OpenAI relationship, Microsoft co-invested alongside Nvidia in Anthropic’s November 2025 round ($5 billion), while Anthropic committed $30 billion to Azure compute in return, deepening a three-way relationship that didn’t exist in our last update.

Google: Continues balancing its own Anthropic investment with broader AI infrastructure bets, and remains notable as the source of the TPU training capacity that both Gemini and portions of Claude’s training run on, an alternative to Nvidia silicon that Huang has publicly acknowledged as the one meaningful defection at scale.

Amazon: Invested $50 billion into OpenAI as part of the same February 2026 round that included Nvidia’s $30 billion, tied to roughly 2 gigawatts of AWS Trainium capacity over eight years, alongside its existing, larger commitment to Anthropic.

Meta: Its Scale AI acquihire remains the standout AI-talent deal of the era; more recently Meta has shifted toward monetizing its own infrastructure directly through Meta Compute rather than expanding equity stakes in other labs.

Investment Activity Timeline

• 2022: 16 corporate deals, 1 NVentures deal
• 2024: 54 corporate deals
• 2025 (full year): 67 corporate deals, 30 NVentures deals, largest single commitments including the OpenAI framework and the November Anthropic investment
• 2026 (through Q2): NVentures alone completed roughly 20 deals in the first five months; corporate development-led AI equity investments exceeded $40 billion in the first four months alone, driven largely by the finalized OpenAI and Anthropic checks

Looking Ahead

What’s already changed from our earlier 2026 predictions: We expected continued expansion into Asia, the Middle East, healthcare AI, and robotics, and that’s broadly held, Legora (legal AI), Aidoc (clinical AI), and Generalist AI (robotics) all point that direction. What we didn’t anticipate was Nvidia explicitly stepping back from its two largest AI lab bets just months after finalizing them.

What to watch through the rest of 2026:

• Whether OpenAI’s and Anthropic’s expected IPOs actually arrive this year or slip into 2027, and what that means for Nvidia’s stated rationale for pulling back
• Whether regulatory scrutiny in the US and EU moves from informal attention to formal investigation
• Whether Nvidia’s cloud-infrastructure investments (CoreWeave, Nebius, Nscale, Lambda) face the same reconsideration as its AI-lab equity, or continue expanding unchecked
• Continued acceleration in vertical, inference-heavy application investments as Nvidia positions itself less as a chip supplier and more as, in its own words, an AI infrastructure platform

Conclusion

Nvidia’s 2025-2026 investment activity tells two stories. The first is more of the same, faster pace, bigger checks, continued expansion into cloud infrastructure, quantum computing, and now vertical AI applications like legal and clinical tools. The second is a genuine inflection point: Nvidia finalized its two largest AI lab investments ever, $30 billion into OpenAI and $10 billion into Anthropic, and then almost immediately signaled it was done making that kind of bet, as the two companies’ rivalry turned public and political.

For entrepreneurs and investors, the signal is worth taking seriously. Even the company with the deepest pockets and the clearest strategic interest in a thriving AI ecosystem is drawing a line under mega-investments in frontier labs. What replaces that capital, whether from public markets once the IPOs land, from other strategics, or from a cooling venture environment, is the open question the rest of 2026 will answer.

The post NVIDIA Top Startup Investments in 2026 appeared first on AI Funding Tracker.

Along the way, OpenAI closed a $122 billion funding round at an $852 billion valuation, watched Anthropic overtake that mark in a single quarter, and got asked by the US government to slow-walk its next flagship model over security concerns.

Here’s where things actually stand.

ChatGPT Passes 1 Billion Users, By Two Different Measures

OpenAI told Cannes Lions attendees on June 22, 2026, that ChatGPT has surpassed 900 million weekly active users, roughly double the figure from February 2025. Separately, Sensor Tower data reported by Reuters put the ChatGPT app itself past 1 billion monthly active users in June 2026, the fastest app in history to reach that scale on its own terms.

Growth timeline:

• February 2025: 400 million weekly active users
• July 2025: 700 million weekly active users
• December 2025: 800 million+ weekly active users
• February 2026: 900 million weekly active users, alongside 50 million paying subscribers and 9 million+ paying business seats
• June 2026: ChatGPT app crosses 1 billion monthly active users (Sensor Tower/Reuters)

But market share tells a different story than raw growth. ChatGPT’s web traffic share has fallen sharply even as absolute usage climbs, from around 87% in January 2025 to under 57% by March 2026, according to Similarweb data. Google’s Gemini has been the biggest beneficiary, growing from roughly 5.7% to over 25% of web traffic share in the same period. ChatGPT is still the largest single platform by a wide margin. It’s just no longer the only one people default to.

Revenue reality check: Audited figures reviewed by the Financial Times show OpenAI generated $13.07 billion in revenue during 2025, more than tripling from $3.7 billion the year before, against a reported $21 billion net loss for the year. Annualized revenue run rate hit roughly $25 billion by February 2026 and has been estimated closer to $28-30 billion by May.

GPT-5.6: A Model Launch the US Government Slowed Down

OpenAI’s next flagship model, GPT-5.6, has had an unusual rollout. Chief Scientist Jakub Pachocki confirmed the model in June, and a preview system card was published June 26. But the Office of the National Cyber Director and the Office of Science and Technology Policy reportedly asked OpenAI to phase the release due to security concerns, and Sam Altman has told employees GPT-5.6 access is being granted per-customer rather than opened widely.

Why it matters: This is part of a broader pattern of the US government inserting itself directly into frontier model release timing, something that also touched Anthropic’s Claude Fable 5 and Mythos 5 releases earlier this year. Frontier AI access is increasingly being treated as a national security lever, not just a product decision.

What’s confirmed about GPT-5.6:

• Notable gains in UI generation and agentic coding, reportedly competitive with or ahead of rival coding-focused models
• OpenAI has been shipping new models roughly every seven weeks through 2026
• The Wall Street Journal reported OpenAI is preparing significant API price cuts alongside the launch, a move read as a direct response to Anthropic’s pricing on its own new Fable 5 and Mythos 5 models

Jalapeño: OpenAI Builds Its First Chip

On June 24, 2026, OpenAI and Broadcom unveiled Jalapeño, OpenAI’s first custom-designed AI chip, built specifically for LLM inference rather than adapted from general-purpose accelerators.

What it is:

• A “blank-slate” inference chip co-developed with Broadcom (silicon implementation, networking) and Celestica (board and rack systems)
• Went from design to tape-out in about nine months, roughly half the usual 1.5-2 year ASIC development cycle, partly by using OpenAI’s own models to accelerate chip design work
• Engineering samples are already running workloads in the lab, including GPT-5.3-Codex-Spark, at production target frequency and power
• Initial deployment targeted for late 2026, scaling through 2027 and “really going full tilt” in the first half of 2028, according to Broadcom CEO Hock Tan

Why it matters: Every major AI lab with the balance sheet to do it is now designing its own silicon, Google has TPUs, Amazon has Trainium, Microsoft has Maia, Meta has MTIA. Jalapeño is OpenAI’s answer, aimed squarely at cutting the cost of serving ChatGPT, Codex, and API traffic rather than training new models. It also gives OpenAI’s IPO pitch a concrete answer to “how do you get to profitability”: cheaper inference at the scale ChatGPT now operates at.

The chip sits inside a much larger hardware web. In February 2026, OpenAI closed a $110 billion-plus funding round that included a $30 billion direct investment from Nvidia (tied to 10 gigawatts of Vera Rubin compute) and a $50 billion investment from Amazon (tied to roughly 2 gigawatts of AWS Trainium capacity over eight years). OpenAI also has agreements with AMD for Instinct MI450 GPUs. Nvidia remains central to training workloads even as Jalapeño targets inference.

ChatGPT Ads: OpenAI Calls Itself “an Advertising Business”

The biggest shift in OpenAI’s business model this year has nothing to do with model quality. At its first-ever Cannes Lions appearance on June 22, 2026, Chief Revenue Officer Denise Dresser told the room OpenAI is “clearly in the advertising business now,” a sharp reversal from Sam Altman calling ads “a last resort” in 2024.

What’s live:

• ChatGPT ads launched in testing on February 10, 2026, starting with free-tier and Go-plan ($8/month) users in the US
• Now live in seven markets (US, UK, Canada, Australia, New Zealand, Japan, South Korea), with Brazil and Mexico next
• The only ad format is a “chat card,” a sponsored box with title, description, image, and link, shown below the model’s organic answer
• Paid Plus ($20/month) and Pro ($200/month) tiers remain permanently ad-free
• A self-serve Ads Manager opened May 5, 2026, with no minimum spend, supporting both CPM and CPC bidding

Early numbers:

• The pilot crossed $100 million in annualized revenue within about ten weeks, with under 600 advertisers at that point
• Over 2,000 brands are now advertising via Criteo alone as of June 2026
• Initial CPMs started around $60 and fell to roughly $25 within ten weeks as the ad pool scaled, with many advertisers shifting to cost-per-click bidding
• OpenAI says roughly 20% of ChatGPT queries carry direct commercial intent, with travel, retail, health, beauty, and financial services as the strongest early categories
• Ad close-out (dismissal) rates reportedly fell 50% since February, which OpenAI points to as evidence of improving ad relevance

The projections are steep: OpenAI has told advertisers to expect $2.4-2.5 billion in ad revenue for 2026, rising to $11 billion by 2027, and $100 billion by 2030. Hitting that 2030 number would require capturing roughly a tenth of global digital ad spend and would put OpenAI’s ad business at around half of Meta’s current annual ad revenue, on a product that’s barely five months old.

Why it matters for marketers: For the first time, there’s a credible “third gatekeeper” alongside Google and Meta for intent-driven advertising. But the measurement and attribution infrastructure both incumbents spent a decade building doesn’t exist yet for ChatGPT. OpenAI has moved fast on partnerships (LiveRamp, Criteo, Adobe, StackAdapt) to close that gap.

The IPO: $1 Trillion or Bust, Possibly Until 2027

OpenAI confidentially filed a draft S-1 with the SEC on June 8, 2026, a week after Anthropic did the same. For a moment it looked like both companies were racing toward late-2026 listings, following SpaceX’s public debut on June 12.

Then the timeline slipped. The New York Times reported June 26 that OpenAI is leaning toward pushing its IPO into 2027 rather than list below its $1 trillion valuation target. Advisers reportedly gave Sam Altman two options: list sooner at a lower valuation, or wait until 2027 to hold out for $1 trillion. Altman called cutting the target a “nonstarter.”

Why the hesitation:

• CFO Sarah Friar has pushed for a 2027 timeline, citing roughly $600 billion in future infrastructure spending commitments through 2030 and the difficulty of meeting public-company reporting standards on a compressed schedule
• SpaceX’s volatile post-IPO trading has reportedly made OpenAI’s advisers more cautious about retail investor appetite for richly valued tech listings
• Amazon’s $35 billion investment commitment only fully unlocks once OpenAI goes public or reaches AGI, which puts some financial pressure against delaying too long
• OpenAI’s own confidential filing included language stating timing is undecided and “it may be a while,” which some read as the company keeping optionality rather than being forced to delay

Where the valuation stands: OpenAI’s last private mark is $852 billion, set when its $122 billion round closed in March 2026 with anchor investments from Amazon, Nvidia, and SoftBank. Anthropic’s $65 billion Series H in late May pushed its own valuation to $965 billion, overtaking OpenAI in private markets for the first time. Prediction markets have reacted accordingly: as of late June, odds of an OpenAI IPO by the end of 2026 sat around 24%.

Competition: A Three-Way Race, Not a Solo Lead

Google (Gemini):

• Gemini app grew from roughly 400 million to over 900 million monthly active users between May 2025 and Google I/O in May 2026
• AI Mode inside Search now claims over 1 billion monthly users on its own
• Web traffic share has climbed past 25%, more than tripling in a year, the fastest-growing major AI platform by that measure

Anthropic (Claude):

• Overtook OpenAI’s private valuation in May 2026 with a $65 billion Series H at a $965 billion post-money valuation
• Run-rate revenue crossed $47 billion in May 2026, up from $14 billion at its February round, and the company projects it will turn its first operating profit in Q2 2026
• Filed its own confidential S-1 on June 1, 2026, a week ahead of OpenAI
• Released Fable 5 and Mythos 5 in June 2026, though both were briefly suspended to comply with US export controls before access was restored July 1

Meta AI:

• Still cites 1 billion monthly active users, a figure it hasn’t refreshed since May 2025 even as ChatGPT and Gemini both report newer, larger numbers
• Shipped its own flagship model, Muse Spark, in April 2026, and separately entered the cloud infrastructure business (Meta Compute) on July 1

Where OpenAI still leads:

1. Raw scale, largest single AI platform by weekly and monthly users
2. Revenue diversity, now running consumer subscriptions, enterprise seats, API access, and a fast-scaling ads business simultaneously
3. Infrastructure ambition, the only lab besides the hyperscalers themselves now designing its own inference silicon

Where OpenAI is under real pressure:

1. Market share erosion, losing web traffic share to Gemini faster than its absolute user growth is adding new ground
2. Valuation gap, Anthropic now holds a higher private valuation, a genuine reversal after years of OpenAI being the default leader
3. Regulatory friction, government-mandated phased rollout of GPT-5.6 signals frontier model access is no longer purely a company decision
4. Burn rate, a reported $21 billion net loss in 2025 against $600 billion in future infrastructure commitments keeps the “path to profitability” question open

The Bottom Line

OpenAI heads into the second half of 2026 bigger than ever and more exposed than ever, at the same time. ChatGPT has genuinely become a mass-market product, past a billion users by more than one measure, embedded in how hundreds of millions of people research, shop, and work. The ad business, four months old, is already generating real revenue and forcing marketers to think about a third major ad platform.

Results so far:

• 1 billion+ monthly active app users (Sensor Tower/Reuters), 900 million+ weekly actives (OpenAI)
• $13 billion in 2025 revenue, tripling year over year, with run rate estimated near $28-30 billion by mid-2026
• ChatGPT ads live in seven countries, on pace for $2.4-2.5 billion in 2026 revenue
• OpenAI’s first custom chip, Jalapeño, already running lab workloads ahead of a late-2026 deployment

The risks:

• A reported $21 billion net loss in 2025 and $600 billion in infrastructure spending commitments through 2030
• Web traffic market share sliding even as absolute usage grows, with Gemini and Claude both gaining faster in percentage terms
• The IPO, once expected by late 2026, now more likely pushed into 2027 as Altman holds out for a $1 trillion valuation
• Government-imposed limits on GPT-5.6’s rollout, a new kind of constraint frontier labs haven’t had to plan around before

OpenAI built the product that started this entire race. Whether it can convert a billion users and a trillion-dollar ambition into a business that actually turns a profit, on its own timeline rather than the market’s or Washington’s, is the story that will define the rest of 2026.

The post ChatGPT News 2026: 1 Billion Users, OpenAI’s First Chip & the IPO That Keeps Slipping appeared first on AI Funding Tracker.

The bigger story since our last update: Meta finally shipped its long-delayed flagship model, now called Muse Spark (internally, Avocado), after three deadline slips. Meta also expanded its Nvidia partnership into a multi-year, multi-gigawatt chip deal, and on July 1, 2026, it stunned Wall Street by announcing a new cloud business, Meta Compute, to sell its excess AI infrastructure. Chip stocks tumbled and Meta’s own share price climbed above $600 for the first time.

Meta isn’t just spending on AI anymore. It’s trying to turn that spending into a second business.

1 Billion Users: Still Meta’s Headline Number, But Rivals Are Catching Up

Meta AI hit 1 billion monthly active users in Q1 2025, CEO Mark Zuckerberg announced at Meta’s May 2025 shareholder meeting, at the time the fastest an AI platform had reached that scale.

Growth timeline (unchanged since the milestone):

• September 2023: Beta launch (limited markets)
• Late 2024: 500M monthly active users
• Q1 2025: 700M monthly active users
• May 2025: Crossed 1B monthly active users
• Since then: Meta has not disclosed an updated Meta AI MAU figure

Why that gap matters now: In June 2026, ChatGPT’s app crossed 1 billion global monthly active users too, according to Sensor Tower estimates reported by Reuters, the fastest app in history to reach that scale on its own metric. Google’s Gemini app, meanwhile, grew from roughly 400 million monthly users a year earlier to over 900 million by Google I/O in May 2026, and Google says AI Mode inside Search now reaches more than 1 billion monthly users on its own. Anthropic’s Claude, while still far smaller in absolute users, nearly tripled its global web traffic share in a single quarter (from about 2.2% in December 2025 to roughly 6% by March 2026).

Meta’s distribution advantage, AI built into WhatsApp, Instagram, Facebook, and Messenger, hasn’t disappeared. But the “fastest growing AI platform” story now has three or four contenders, not one.

Platform breakdown (last disclosed figures, Meta has not refreshed these):

• WhatsApp: ~630M Meta AI users, ~63% of interactions
• Instagram: ~270M Meta AI users, ~27% of interactions
• Facebook: ~100M Meta AI users, ~10% of interactions
• Standalone Meta AI app: 10-20M users, expanding beyond its original US/Canada/Australia/New Zealand markets

Geographic distribution: India still leads globally, driven by WhatsApp’s 500M+ Indian user base. Global reach sits at 60+ countries and 13+ languages.

Muse Spark Has Launched: Avocado’s Public Debut, Three Delays Later

The model long known internally as Avocado didn’t arrive on the original late-2025 timeline, or the March 2026 revision, or even the May window. It finally launched publicly on April 8, 2026, under a new name: Muse Spark, the first model out of Meta Superintelligence Labs (MSL) under chief AI officer Alexandr Wang.

What shipped:

• Muse Spark now powers the standalone Meta AI app and meta.ai, with rollout planned across Facebook, Instagram, and WhatsApp
• It accepts voice, text, and image input, but produces text only output. Despite Meta’s earlier “multimodal” framing, this version doesn’t generate images or video
• It runs in tiered modes: an instant mode for casual queries and a thinking mode for step by step reasoning tasks
• A standout feature is “visual coding,” generating interactive websites, dashboards, and mini-games from a single prompt
• It can spin up multiple sub-agents in parallel (for example, splitting a trip-planning request across itinerary, destination, and activity agents)
• All versions are free to use, with Meta planning to release an open-source variant alongside the proprietary one

How it performs: A Meta executive told Axios the model isn’t a new state of the art but is competitive with leading labs on tasks like multimodal understanding and health-related queries, scoring 89.5% on GPQA Diamond and leading rivals on HealthBench Hard at 42.8%. Meta has openly acknowledged a persisting gap in coding performance against models like GPT-5.4.

Why the three delays happened: Internal testing through Q1 2026 reportedly placed Avocado’s benchmark performance somewhere between Google’s Gemini 2.5 and Gemini 3.0, solid progress over Llama 4, but short of the frontier bar Meta needed to hit. Meta even explored temporarily licensing Google’s Gemini to power its own products while Avocado caught up, though no such deal materialized.

Mango: still unreleased. Unlike Avocado/Muse Spark, Meta’s image and video model codenamed Mango has not had a public launch as of this update. It remains in development inside MSL, aimed at competing with OpenAI’s Sora and Google’s Veo, with no confirmed release date.

Nvidia Mega-Deal: Now Public, Still Undisclosed in Dollar Terms

Meta’s expanded Nvidia partnership, announced February 17, 2026, is now fully underway.

What’s confirmed:

• Millions of Blackwell and upcoming Rubin GPUs
• Standalone Grace CPUs, Nvidia’s first large-scale standalone Grace deployment globally
• Vera CPUs slated for 2027
• Spectrum-X Ethernet networking and security tooling for WhatsApp AI features

Deal size: Neither company has disclosed the dollar figure. Analyst back-of-envelope math (roughly $3.5M per GPU-equipped rack) puts a million-GPU tranche in the ballpark of $48 billion, consistent with earlier “tens of billions” estimates.

Multi-vendor hedge continues: Meta is also evaluating Google TPUs for 2027 deployment alongside AWS Trainium, AMD GPUs, and in-house silicon, though Nvidia remains its primary supplier.

Data Centers: Hyperion and Prometheus Move From Blueprint to Buildout

Meta’s flagship sites are no longer just announcements. They’re under active construction, with clearer specs than we had in February.

• Prometheus (New Albany, Ohio): 1-gigawatt facility, tied directly to the new Nvidia chip allocation; reports point to a Meta-linked land purchase nearby suggesting further campus expansion
• Hyperion (Richland Parish, Louisiana): Now specified at 1.5 gigawatts by late 2027, scaling to 5 gigawatts by 2030, designed to house 1.3 million+ GPUs. Powered by a mix of natural gas, renewables, battery storage, and nuclear baseload, with direct liquid cooling for high-density racks
• March 27, 2026: Entergy Louisiana signed a new agreement tied to Hyperion, promising customers an additional $2 billion in savings over 20 years, a response to local concern about the data center’s power draw

2026 capex guidance: Still $115-135 billion, as set on the January 2026 earnings call. Actual Q1 2026 capex came in lower than Wall Street expected, at $19.84 billion against a $27.57 billion estimate, a gap Meta hasn’t fully explained but one that briefly rattled investors around the earnings release.

New: Meta Compute, Meta Enters the Cloud Business

The biggest development since our last update, and one that wasn’t on anyone’s radar in February: on July 1, 2026, Bloomberg reported Meta is building Meta Compute, a cloud infrastructure unit that will sell spare AI computing power to outside customers, putting Meta in direct competition with AWS, Microsoft Azure, and Google Cloud.

Why now: Meta has spent years defending its AI capex by tying it to advertising gains. But free cash flow pressure has grown alongside that spending, with capex going from $37.2B in 2024 to $69.6B in 2025 and toward $135B at the midpoint for 2026. Selling excess capacity reframes years of “unexplained” spending as a hedge, not just a bet.

Two possible shapes, according to Bloomberg’s reporting: a “bare metal” model where customers bring their own software (similar to how SpaceX rents capacity to firms like Anthropic), or a fuller cloud platform with developer tools layered on top, closer to AWS Bedrock.

Market reaction was immediate and sharp:

• Meta’s own stock jumped nearly 9%, pushing shares above $600
• Micron fell more than 10%; SanDisk dropped over 14%; Intel and AMD each lost 7-10%
• Nvidia dipped only slightly (about 1.25%), standing out against the broader rout
• CoreWeave and Nebius, both GPU-rental specialists, fell 14% and 17% respectively on fears Meta would undercut their pricing
• The sell-off spread to Asian chipmakers, with Samsung and SK Hynix both down more than 7%

The signal: for the first time, a hyperscaler openly admitted it has more AI compute than it needs right now, flipping years of assumed scarcity into a supply question. Whether Meta can actually run a cloud business (support, reliability, enterprise sales) as well as it runs social platforms is untested.

AI Revenue: Q1 2026 Numbers Are In

Meta’s Q1 2026 earnings, reported April 29, 2026, gave the clearest read yet on whether AI spending is showing up in the business.

• Revenue: $56.31 billion, up 33% year over year, Meta’s fastest quarterly growth since 2021
• Net income: $26.77 billion
• Ad revenue: $55.02 billion, 98% of total revenue, still Meta’s overwhelming monoculture
• Ad impressions: +19% YoY; average price per ad: +12% YoY, both credited to AI tooling across the ad stack
• Advertisers using Meta’s AI creative tools: 8 million, double the figure from a year earlier
• Daily Active People (DAP): 3.56 billion, a 4% YoY increase but a miss against the 3.62 billion Wall Street expected. Meta attributed the shortfall partly to internet disruptions in Iran and a WhatsApp restriction in Russia
• Click-to-message ads grew more than 50% YoY in the US; paid WhatsApp messaging kept scaling past a $2 billion annual run-rate

Takeaway: AI still hasn’t created a clean new revenue line at Meta, no “Muse Spark subscription” line item exists yet. But it’s visibly lifting the core ads engine, and Meta Compute is now a credible second bet.

Competition: The Race Has Gotten More Crowded, Not Less

OpenAI (ChatGPT):

• Crossed 1 billion monthly active app users in June 2026 (Sensor Tower/Reuters), the fastest app ever to reach that scale
• 900 million weekly active users as of February 2026
• Annualized revenue run rate over $25 billion (February 2026)
• Web traffic share has fallen sharply even as raw usage grows, from roughly 87% in January 2025 to under 57% by March 2026, as competitors take share

Google (Gemini):

• Gemini app grew from ~400M to over 900M monthly active users between May 2025 and May 2026 (Google I/O disclosure)
• AI Mode in Search now claims over 1 billion monthly users on its own
• Web traffic share climbed past 25%, more than tripling in a year
• A reported Apple deal, a custom 1.2-trillion-parameter Gemini model powering next-gen Siri, remains on track for a 2026 rollout with iOS 27

Anthropic (Claude):

• Smaller in absolute users but the fastest growing by percentage: web traffic share roughly tripled in a single quarter, from ~2.2% to ~6% between December 2025 and March 2026
• Has continued shipping frontier models through 2026, keeping pace with Meta’s own Muse Spark on several benchmarks

Where Meta still leads:

1. Distribution: no rival can embed AI across apps used by roughly half the world’s population
2. Ad monetization proof points: 8 million advertisers now using AI creative tools, real revenue impact visible in Q1 numbers
3. Infrastructure scale: among the largest capex commitments in tech, now paired with a plan (Meta Compute) to monetize the excess

Where Meta still trails:

1. Model performance: Muse Spark closed some of the gap but still lags on coding versus GPT-5.4 and top Gemini variants
2. Visual generation: Mango, Meta’s answer to Sora and Veo, still hasn’t shipped
3. User transparency: Meta hasn’t updated its headline Meta AI user count since May 2025, even as ChatGPT and Gemini both report fresh, larger numbers
4. Enterprise trust: privacy history continues to limit adoption for sensitive workloads compared to Claude and ChatGPT

The Bottom Line

Meta’s AI bet is now playing out on three fronts at once: consumer models, infrastructure, and, as of July, a cloud business built to monetize its own excess capacity.

Results so far:

• Muse Spark live and powering the Meta AI app, with an open-source variant planned
• $56.3B in Q1 2026 revenue, up 33% year over year, with AI tooling visibly lifting ad performance
• 8 million advertisers now using Meta’s AI creative tools
• Prometheus and Hyperion moving from blueprints to active construction, backed by a firm Nvidia supply commitment

The risks:

• $115-135B in annual spending with a still-unclear return timeline
• Mango’s absence leaves a real gap against Sora and Veo in the fastest growing part of generative AI
• Meta hasn’t refreshed its headline Meta AI user numbers in over a year, while ChatGPT and Gemini both report bigger, fresher figures
• Meta Compute is unproven; running a cloud service business is a different discipline than running social platforms
• Coding performance remains a weak spot even in Meta’s newest model

Meta is no longer just trying to catch up on model quality. It’s trying to build a business model around the infrastructure it already owns, while still finishing the model race it started. Whether Muse Spark’s next iteration and Mango’s eventual launch close the remaining technical gaps, and whether Meta Compute becomes a durable revenue stream rather than a headline, will shape how the rest of 2026 plays out.

The AI race isn’t over. Meta just added a second front to it.

The post Meta AI News 2026: Muse Spark Launches, Nvidia Mega-Deal Expands & Meta Enters the Cloud Business appeared first on AI Funding Tracker.

This article breaks down exactly who owns OpenAI in 2026, how much each stakeholder holds, what the October 2025 restructuring did, and how the giant March 2026 fundraise and a June 2026 IPO filing reshaped the picture.

Key Takeaways on OpenAI Ownership

• OpenAI is now valued at $852 billion, following a record $122 billion round that closed in March 2026.
• Microsoft remains the largest external shareholder (~27% at the October 2025 recap), but its position was diluted by the 2026 round and its commercial deal was loosened in April 2026.
• The nonprofit still runs the show. Despite holding ~26%, the OpenAI Foundation appoints the entire PBC board and keeps mission control.
• Amazon, SoftBank, and Nvidia anchored the 2026 round with Amazon’s entry, notable given it’s also Anthropic’s biggest backer.
• The Microsoft partnership changed. Revenue-share payments are now capped, Azure exclusivity is gone, and Microsoft keeps resell rights through 2032.
• . Sam Altman still owns no equity (pending any future grant), an unusual choice for a founder of a company this large.
• The Musk lawsuit is over; Musk lost in May 2026, removing a major legal overhang.
• The IPO is here. OpenAI filed confidentially on June 8, 2026, targeting a $1 trillion-plus valuation, with a listing possible as early as September 2026.

Current OpenAI Ownership Structure (2026)

OpenAI’s most recent disclosed ownership breakdown comes from its October 2025 recapitalization, when the company was valued at $500 billion. That snapshot looked like this:

• Microsoft: 27% (~$135 billion at the time)
• OpenAI Foundation (nonprofit): 26% (~$130 billion)
• Current and former employees: 25% (~$125 billion)
• 2025 fundraise investors: 13% (~$65 billion)
• 2024 fundraise investors: 4% (~$20 billion)
• IO shareholders: 2% (~$10 billion)
• Original early investors: 1% (~$5 billion)

Important: This is the October 2025 picture. In March 2026, OpenAI closed a $122 billion round at an $852 billion valuation, which diluted every existing holder and brought in major new investors. OpenAI has not re-published a clean, post-round percentage table, so treat the figures above as the last fully disclosed snapshot, the dollar values in particular are now higher, and the percentages are lower, than they were at $500 billion.

The clearest example: Microsoft’s stake was pegged at $135 billion (27%) in October 2025, but by May 2026 — at the higher valuation — Microsoft’s holdings tied to its OpenAI commitments were reported to be worth roughly $228 billion.

The March 2026 Round: $122 Billion at $852 Billion

This is the development that reshaped OpenAI’s cap table. On March 31, 2026, OpenAI closed its largest-ever funding round: $122 billion in committed capital at an $852 billion post-money valuation, up from the $110 billion the company had first announced in February.

The lead participants:

• Amazon: up to $50 billion (a notable entry – Amazon is also the largest backer of OpenAI’s rival, Anthropic)
• SoftBank: $30 billion
• Nvidia: $30 billion
• Microsoft: continued participation (amount undisclosed)
• Andreessen Horowitz and D. E. Shaw Ventures: co-participants

For the first time, OpenAI also opened the round to individual investors through bank channels, raising about $3 billion from retail participants. The round cemented OpenAI as the most valuable private company in history at the time, a title it held until Anthropic passed it at a $965 billion valuation in late May 2026.

How OpenAI’s Structure Changed in October 2025

OpenAI made a massive shift on October 28, 2025, completing a recapitalization that changed its corporate structure.

The old structure: OpenAI operated with a nonprofit parent controlling a “capped-profit” for-profit subsidiary. Investors could earn up to 100x returns, after which all profits reverted to the nonprofit.

The new structure: The OpenAI Foundation (the nonprofit) now controls a Public Benefit Corporation (PBC) called OpenAI Group PBC. The Foundation holds 26% equity but retains full control through governance rights.

Why did this happen? OpenAI needed to raise enormous amounts of capital. The capped-profit model worked when OpenAI looked like the only major AGI effort, but with competitors like Anthropic, xAI, and Google DeepMind, OpenAI needed a simpler capital structure.

The new PBC structure is the same model Anthropic and xAI use. It lets OpenAI attract traditional investors while maintaining mission focus, with the Foundation’s board appointing all members of the for-profit board.

Microsoft’s Stake — and the Renegotiated Deal

Microsoft is OpenAI’s largest external shareholder, having invested more than $13 billion since 2019.

Investment timeline:

• 2019: $1 billion initial investment
• 2023: $10 billion additional investment
• October 2025: Microsoft’s stake disclosed at ~27% (~$135 billion) on an as-converted diluted basis
• 2026: Continued participation in the $122 billion round; holdings later valued near $228 billion at the higher valuation

The big change — April 2026: OpenAI and Microsoft announced a revamped partnership agreement. The key shifts:

• OpenAI can now cap its revenue-share payments to Microsoft (reported at roughly $38 billion through 2030, a sharp reduction from the prior projected trajectory)
• OpenAI is now free to serve customers across any cloud provider — Azure cloud exclusivity is gone
• Microsoft retains resell rights for OpenAI technology through 2032
• The revenue-share arrangement still winds down once AGI is independently verified

This was a meaningful loosening of what had been one of the tightest partnerships in tech. The relationship has shown public strain — the renegotiation was announced the same week jury selection began in Elon Musk’s lawsuit against OpenAI and Microsoft (more on that below).

The OpenAI Foundation’s Stake

The OpenAI Foundation (formerly the OpenAI Nonprofit) held 26% equity at the October 2025 recap. What makes it unique: despite that minority stake, the Foundation controls OpenAI Group PBC’s decision-making.

How? The Foundation appoints the entire board of the for-profit PBC. All board members serve on both the Foundation board and the PBC board, keeping OpenAI’s mission — developing safe AGI that benefits humanity — at the center of governance.

The Foundation announced an initial $25 billion commitment toward health breakthroughs and technical solutions for AI resilience, signaling the nonprofit arm has real resources, not just symbolic authority.

Employee Ownership

Current and former OpenAI employees collectively held about 25% at the October 2025 recap. Unusually, OpenAI includes former employees in the ownership pool rather than clawing back unvested equity — building long-term loyalty and keeping alumni invested in the company’s success.

In late 2025, OpenAI authorized a $10.3 billion secondary share sale; current and former employees sold roughly $6.6 billion of stock to investors including SoftBank, Thrive Capital, and Dragoneer. The lower-than-authorized participation was read as employee confidence in long-term value.

Sam Altman, OpenAI’s CEO and co-founder, has publicly stated he holds no equity in the company, remarkable for a founder leading a company of this scale. (Note: there has been periodic reporting about OpenAI considering an equity grant for Altman; as of this update no such grant has been confirmed.)

Major Investors: Amazon, SoftBank, Nvidia, and More

The 2025 and 2026 rounds brought in a deep bench of major investors.

Amazon: Entered OpenAI’s cap table in the 2026 mega-round with a commitment of up to $50 billion — a striking move given Amazon is simultaneously the largest investor in OpenAI’s chief rival, Anthropic.

SoftBank: The Japanese investment giant committed $30 billion in the 2026 round, on top of its earlier $40 billion commitment that completed in late 2025. SoftBank remains one of OpenAI’s most significant capital partners.

Nvidia: Committed $30 billion in the 2026 round, deepening the chipmaker’s financial ties to its largest AI customer.

Thrive Capital: Led by Josh Kushner, a consistent OpenAI backer since 2023, participating across multiple rounds and the secondary sale.

Other major investors: Dragoneer, Coatue, Altimeter Capital, MGX (Abu Dhabi sovereign wealth fund), T. Rowe Price, Andreessen Horowitz, and D. E. Shaw Ventures.

Early investors (1% collective): The original 2019 backers — Khosla Ventures, Reid Hoffman Foundation, Y Combinator, Paul Buchheit, and the University of Michigan — invested $194 million in 2019; their 1% stake was worth roughly $5 billion at the October 2025 valuation.

Who Founded OpenAI?

OpenAI was founded in December 2015 by a group of tech leaders:

Original founders: Sam Altman (co-chair, now CEO), Elon Musk (co-chair), Greg Brockman (CTO), Ilya Sutskever (Chief Scientist), Wojciech Zaremba, John Schulman, Andrej Karpathy, Trevor Blackwell, Vicki Cheung, Durk Kingma, and Pamela Vagata.

Initial funding pledges: Elon Musk, Peter Thiel, Reid Hoffman, Jessica Livingston, Amazon Web Services, Infosys, and YC Research.

The founding group pledged $1 billion to develop AI “for the benefit of humanity broadly,” with a mission focused on ensuring AGI wouldn’t be controlled by a small group of corporations or governments. OpenAI began as a pure nonprofit; in 2019, facing massive compute costs, it restructured to add a capped-profit subsidiary.

Elon Musk’s Exit and the Lawsuit He Lost

Elon Musk’s relationship with OpenAI became one of tech’s most contentious stories — and in 2026, it reached a courtroom conclusion.

2015–2018: Musk co-founded OpenAI and was deeply involved in early strategy. He pledged $1 billion but, according to OpenAI, contributed roughly $38–45 million before internal conflicts over equity, control, and a proposed Tesla merger led to his departure from the board in February 2018.

2023: Musk launched xAI, a direct competitor.

2024: Musk sued OpenAI and Sam Altman, alleging they abandoned the founding nonprofit mission. OpenAI countersued, calling his tactics “bad faith.”

February 2025: A Musk-led consortium made a $97.4 billion unsolicited bid for the nonprofit controlling OpenAI. OpenAI rejected it, stating the company was “not for sale.”

April–May 2026 — the trial: The case went to a jury before US District Judge Yvonne Gonzalez Rogers in Oakland. Musk sought damages reported in the range of $79–134 billion, arguing his early contributions generated a large share of OpenAI’s value. Microsoft CEO Satya Nadella and OpenAI co-founder Ilya Sutskever testified. In May 2026, the jury and court ruled against Musk — he lost the case. With the lawsuit resolved, a significant overhang on OpenAI’s IPO plans was removed.

What Happens When OpenAI Achieves AGI?

OpenAI’s structure still includes provisions for Artificial General Intelligence. When OpenAI claims AGI, an independent expert panel must verify it, and Microsoft’s revenue-share rights wind down once AGI is verified.

The April 2026 renegotiation added a wrinkle: even before any AGI determination, OpenAI’s revenue-share payments to Microsoft are now capped (reported at ~$38 billion through 2030). After AGI verification, Microsoft can pursue AGI independently or with third parties, and OpenAI can jointly develop products outside Microsoft’s commercial rights.

The Foundation’s control of the board is designed to keep AGI decisions aligned with safety and broad benefit over pure profit — at least in theory.

OpenAI’s Path to IPO

OpenAI has moved from “preparing” to “filed.” On June 8, 2026, OpenAI confidentially filed for an IPO with the SEC, targeting a valuation exceeding $1 trillion, with Goldman Sachs and Morgan Stanley leading the process. Some reports point to a possible listing as early as September 2026, though OpenAI has signaled timing depends on market conditions and could slip.

The October 2025 recapitalization specifically prepared OpenAI for this path, and the resolution of the Musk lawsuit cleared a major legal overhang. OpenAI’s rival Anthropic filed its own confidential IPO paperwork a week earlier, on June 1 — turning the two-horse AI race into a public-markets race as well.

At a $1 trillion-plus valuation, OpenAI’s listing would be among the largest in history. The company has committed to massive infrastructure spending — including the Stargate initiative with Oracle and SoftBank, and chip agreements with Nvidia, AMD, and Broadcom — and reported generating roughly $2 billion in revenue per month as of early 2026.

What This Means for OpenAI’s Future

OpenAI’s ownership structure reflects a persistent tension: maximizing commercial success while staying true to a nonprofit mission. The PBC structure and a diverse, deep-pocketed investor base — now including Amazon, SoftBank, and Nvidia — provide the hundreds of billions needed to chase AGI.

But maintaining mission focus at an $852 billion valuation, heading into a $1 trillion public listing, creates real conflicts. The Foundation’s board control is designed to prevent profit optimization from overriding safety, yet public-market pressure is a different kind of force than private investors. The loosening of the Microsoft deal, the resolution of the Musk suit, and the IPO filing all point in the same direction: OpenAI is clearing the runway for public markets. Whether mission-first governance survives contact with quarterly earnings is the defining question ahead.

The post Who Owns OpenAI? Complete Ownership Breakdown (2026) appeared first on AI Funding Tracker.

But Anthropic’s ownership works very differently from OpenAI’s. There’s no neat, publicly disclosed cap table here. Anthropic is a private Public Benefit Corporation that has never released exact equity percentages, and its control structure deliberately separates “who owns the most shares” from “who actually steers the company.”

This article breaks down what’s actually known about who owns Anthropic in 2026, how much each major stakeholder holds, and why the Long-Term Benefit Trust matters more than any single investor’s check.

Current Anthropic Ownership Structure (2026)

Unlike OpenAI, Anthropic has never published a full ownership breakdown. What we know comes from funding announcements, court filings, and investor disclosures. Here’s the most accurate picture as of June 2026:

• Founders & employees: The largest collective ownership block. The seven co-founders plus current and former employees hold the majority of equity, though exact percentages aren’t disclosed.
• Amazon: Largest single outside investor. ~$8 billion invested, with a commitment of up to $25 billion more. Estimated stake in the mid-teens percent range.
• Google (Alphabet): 14% per 2025 court filings, contractually capped at 15%, with no voting rights or board seats.
• Nvidia: Committed up to $10 billion (November 2025); estimated low-single-digit stake.
• Microsoft: Committed up to $5 billion (November 2025); low-single-digit stake.
• Institutional & VC investors: ICONIQ, Lightspeed, Sequoia, GIC, Coatue, Fidelity, Menlo Ventures, Spark Capital, Salesforce Ventures (~$5B stake), MGX, Qatar Investment Authority, Altimeter, Dragoneer, Greenoaks, and dozens more.

The company is valued at $965 billion following its May 2026 Series H round. Anthropic confidentially filed for an IPO on June 1, 2026, which means the precise cap table may finally become public when the S-1 is unsealed.

A key caveat: Anyone publishing exact founder or employee percentages for Anthropic is estimating. The company has 94+ investors across 18 funding rounds, and the only externally verified outside-investor figure is Google’s 14% from a 2025 antitrust court filing.

Anthropic’s Corporate Structure: PBC + Long-Term Benefit Trust

This is the part that makes Anthropic genuinely different from most startups.

Anthropic is organized as a Delaware Public Benefit Corporation (PBC). That means its directors are legally required to balance shareholder returns against its public mission of developing safe AI, they can’t be sued purely for leaving money on the table in service of safety.

Sitting on top of that is the Long-Term Benefit Trust (LTBT), designed by Anthropic’s founders with outside counsel from Wilson Sonsini. The Trust is a five-member independent body whose members hold no financial stake in Anthropic. Over time, the LTBT gains the power to elect and remove a majority of Anthropic’s board of directors.

The logic: even if commercial investors own enormous amounts of equity, they can’t simply take control of the board and override the safety mission. The Trust has already used this power—appointing directors like Netflix co-founder Reed Hastings and former Novartis CEO Vas Narasimhan.

The failsafe (and the catch): Per Anthropic’s incorporation documents, a supermajority of shareholders can amend the rules governing the LTBT without the Trust members’ consent. Anthropic describes this as a safeguard in case the structure turns out to be flawed. Critics note it theoretically leaves a door open for large shareholders to force governance changes—though Anthropic maintains the practical hurdles make this near-impossible.

This is the opposite of a typical cap table story. At Anthropic, control is not purely a function of who owns the most shares.

Who Founded Anthropic?

Anthropic was incorporated in January 2021 by a group of senior researchers and executives who left OpenAI over disagreements about AI safety and the pace of commercialization. Anthropic publicly identifies seven co-founders:

• Dario Amodei (CEO) — former VP of Research at OpenAI, led GPT-2 and GPT-3 development, co-invented RLHF
• Daniela Amodei (President) — Dario’s sister, former VP of Operations at OpenAI, previously at Stripe
• Tom Brown — lead author on the GPT-3 paper
• Jack Clark — AI policy lead, head of policy
• Jared Kaplan — theoretical physicist, chief science officer
• Sam McCandlish — research lead, scaling laws
• Christopher Olah — interpretability research pioneer

(Some sources also count Ben Mann among the founding group, which is why you’ll occasionally see “eight co-founders.”)

All seven became billionaires in March 2025 when a $3.5 billion round valued Anthropic at $61.5 billion. In January 2026, the co-founders collectively pledged to donate 80% of their personal wealth to philanthropy aimed at addressing AI-driven inequality—a pledge worth more than $21 billion at the time, announced alongside Dario Amodei’s essay “The Adolescence of Technology.” Forbes estimated Dario’s personal net worth at roughly $7 billion in early 2026.

Amazon: Anthropic’s Largest Outside Investor

Amazon is the biggest single external backer of Anthropic, but it took a winding path to get there.

Investment timeline:

• September 2023: Initial $1.25 billion investment
• March 2024: Additional $2.75 billion, bringing the total to $4 billion
• November 2024: Topped up to a total of $8 billion invested
• April 2026: Agreed to invest up to $25 billion more as part of an expanded infrastructure deal

That latest agreement could eventually push Amazon’s total commitment past $30 billion. In exchange, Anthropic committed to spending more than $100 billion on AWS technologies over 10 years, including Amazon’s custom Trainium AI chips (securing up to 5 gigawatts of compute capacity).

Amazon’s exact stake isn’t disclosed, but most analysts estimate a mid-teens percentage. The position is already enormously profitable on paper: in Q1 2026 alone, Amazon booked $16.8 billion in pre-tax gains from its Anthropic holding, including $12.3 billion from revaluing the stake after the Series G round.

Crucially, like Google, Amazon’s investment comes with limited control—no governance veto over Anthropic’s safety mission.

Google: 14% and Capped

Google (Alphabet) is the one outside investor whose exact stake is publicly verified. Thanks to legal filings Anthropic submitted in a Google antitrust case, accidentally sent to The New York Times in unredacted form, we know that:

• Google owns 14% of Anthropic
• Google is contractually capped at owning no more than 15%
• Google holds no voting rights, no board seats, and no board observer rights

Google has committed substantial capital and cloud credits over time (reported commitments range up to tens of billions in combined cash and Google Cloud compute). The arrangement reflects Anthropic’s deliberate dual-cloud strategy: it runs workloads on both AWS and Google Cloud to avoid dependence on any single provider.

In a notable 2025 development, the U.S. Justice Department dropped a proposal that would have forced Google to divest its stakes in generative AI companies—allowing Google’s 14% in Anthropic to remain intact.

Microsoft and Nvidia Join In (2025)

In a surprising November 2025 move, two more giants entered Anthropic’s cap table:

• Nvidia committed up to $10 billion
• Microsoft committed up to $5 billion—despite already owning 27% of Anthropic’s rival OpenAI

The combined $15 billion deal tied Anthropic closer to both companies commercially. Anthropic agreed to purchase $30 billion in Microsoft Azure cloud capacity and to adopt Nvidia’s latest chip architectures. As Microsoft CEO Satya Nadella put it, the companies are “increasingly going to be customers of each other.”

Both stakes are believed to be in the low-single-digit percentage range.

Anthropic’s Funding History and Valuation Climb

Few companies in history have repriced this fast. Here’s the trajectory:

• 2021: Founded; early backing included a $580M round (with now-infamous FTX/Alameda involvement, later divested via bankruptcy proceedings)
• 2023: Google, Spark Capital, and others invest; Amazon partnership begins
• March 2025: $3.5 billion round at $61.5 billion (led by Lightspeed)
• September 2025: Series F, $13 billion at $183 billion (led by ICONIQ)
• February 2026: Series G, $30 billion at $380 billion (led by GIC and Coatue)
• May 2026: Series H, $65 billion at $965 billion (co-led by Altimeter, Dragoneer, Greenoaks, Sequoia, Capital Group, Coatue, D1, GIC, ICONIQ, and XN)

In total, Anthropic has raised roughly $125–132 billion across 18 funding rounds. The Series H is described as likely its last private raise before going public.

The valuation jump is backed by real revenue: Anthropic’s run-rate revenue crossed $47 billion by May 2026, up from $30 billion earlier in the year and just $10 billion in annual revenue the year before—driven largely by its AI coding assistant, Claude Code.

Employee Ownership and the 2026 Tender Offer

Anthropic’s roughly 2,500 employees hold a significant collective equity stake, though the company doesn’t break out the exact figure.

In February 2026, alongside its primary fundraise, Anthropic ran an employee tender offer of $5–6 billion, letting current and former employees sell shares to investors at a roughly $350 billion valuation. This gave early employees liquidity without waiting for the IPO—a common pattern at late-stage AI labs, and a sign of how much paper wealth has accumulated inside the company.

Anthropic’s Board of Directors

Because of the PBC and LTBT structure, Anthropic’s board is a mix of founders, independent directors appointed by the Trust, and investor-linked seats. As of 2026 it includes:

• Dario Amodei (CEO, co-founder)
• Daniela Amodei (President, co-founder)
• Reed Hastings (Netflix co-founder; appointed by the LTBT)
• Vas Narasimhan (former Novartis CEO; appointed by the LTBT in April 2026)
• Chris Liddell (former Microsoft and General Motors CFO)
• Yasmin Razavi (Spark Capital)

Note that the major cloud investors—Amazon and Google—do not hold board seats, a deliberate feature of Anthropic’s governance design.

The Road to IPO

Anthropic confidentially filed a draft S-1 with the SEC on June 1, 2026, kicking off what could become one of the largest public offerings in history. At its $965 billion private valuation, the company is approaching the trillion-dollar mark.

The IPO is significant for ownership transparency reasons too: when the prospectus is eventually made public, it should reveal—for the first time, the actual cap table, including precise founder, employee, and investor percentages that have been estimated until now.

Anthropic is racing rival OpenAI (valued at $852 billion in March 2026) to the public markets, alongside other 2026 AI-era listings. How the company preserves its Long-Term Benefit Trust governance as a public company will be one of the most closely watched questions in tech.

Key Takeaways on Anthropic Ownership

Here’s what matters about who owns Anthropic in 2026:

1. No single owner controls Anthropic. Ownership is split among founders, employees, cloud giants, and dozens of institutional investors—with no majority holder.

2. The founders and employees hold the biggest collective stake, but exact percentages remain undisclosed until the IPO prospectus is unsealed.

3. Amazon is the largest outside investor at ~$8 billion (with up to $25 billion more committed), holding an estimated mid-teens percentage.

4. Google owns exactly 14%—the only verified outside figure—capped at 15%, with no voting rights or board seats.

5. Governance is separated from ownership. The Long-Term Benefit Trust can elect a majority of the board, so big investors can’t simply buy control of the safety mission.

6. Microsoft and Nvidia joined in late 2025, adding $15 billion combined and binding Anthropic to Azure and Nvidia hardware.

7. The valuation went vertical—from $61.5 billion in March 2025 to $965 billion in May 2026, on the back of $47 billion in run-rate revenue.

8. An IPO is imminent. The June 2026 confidential filing should finally make Anthropic’s full cap table public.

What This Means for Anthropic’s Future

Anthropic’s ownership structure is a bet that you can raise hundreds of billions in commercial capital without surrendering mission control. The PBC framework and Long-Term Benefit Trust are designed to let Amazon, Google, Microsoft, and Nvidia fund the compute Anthropic needs, while keeping safety-focused governance insulated from pure profit pressure.

Whether that holds at trillion-dollar scale is the open question. Once Anthropic is public, quarterly earnings pressure, activist shareholders, and the LTBT’s “failsafe” amendment clause will all test how durable mission-first governance really is.

For now, Anthropic has pulled off something unusual: it’s simultaneously one of the most heavily funded companies on earth and one whose founders insist no investor can buy the steering wheel.

The post Who Owns Anthropic? Complete Ownership Breakdown (2026) appeared first on AI Funding Tracker.