On February 27, 2026, OpenAI announced it had closed one of the largest private funding rounds in the history of technology – $110 billion at a $730 billion pre-money valuation, drawn from three of the most strategically consequential companies currently operating in the global technology landscape: Amazon at $50 billion, Nvidia at $30 billion, and SoftBank at $30 billion, with the round still open and additional investors expected to join as it progresses.
To appreciate the scale of what happened, consider that venture capitalists invested a total of $170 billion into all U.S. startups across the entirety of 2023 – and OpenAI has now raised nearly two-thirds of that figure in a single transaction, less than four years after ChatGPT launched to a public that had no idea what to do with it. The previous record for a private technology funding round was also OpenAI's own $40 billion raise in March 2025, which had itself shattered Ant Group's $14 billion raise from 2018 — meaning OpenAI has now broken the all-time record for private capital formation twice in under twelve months.
But the headline number, as dramatic as it is, represents only the surface of a far more intricate story. The real substance lies in the structure of the deal: who is investing, on precisely what terms, with what strategic motivations, and what obligations OpenAI has assumed in exchange for receiving this capital.
Who Is Actually Paying, and What They Get Back
None of the three investors in this round should be understood as passive financial backers placing a bet on a promising startup. Each one is making a calculated, deeply strategic commitment tied directly to its own core business interests — and each one has negotiated concrete, operationally meaningful returns that extend well beyond equity ownership.
Amazon: $50 Billion
Amazon's commitment is simultaneously the largest in the round and the most structurally conditional, structured in two tranches with triggers that reveal as much about Amazon's priorities as about OpenAI's financial needs.
| Tranche | Amount | Trigger |
|---|---|---|
| Initial | $15B | Committed now |
| Second | $35B | Conditions met (IPO or other redacted milestones) |
| Expiry | — | December 31, 2028 |
According to filings reviewed by GeekWire, an OpenAI IPO filing is one disclosed trigger that would require Amazon to deploy the remaining $35 billion, while a second set of conditions remains redacted in public documents. Sam Altman has been explicit that these conditions are not AGI-related, dismissing speculation about a triggering mechanism similar to the AGI clause embedded in Microsoft's earlier partnership agreement.
Beyond the equity component, the investment comes wrapped in a commercial arrangement of extraordinary scope:
- AWS becomes the exclusive third-party cloud provider for OpenAI Frontier, the company's new enterprise agent platform
- OpenAI expands its existing $38 billion AWS agreement by an additional $100 billion over eight years
- The two companies will jointly develop customized AI models designed specifically for Amazon's own engineering and customer-facing applications
The commercial logic is important to hold alongside the investment figures, because a significant portion of that $50 billion will flow back to Amazon in the form of cloud revenue — a dynamic that Wall Street analysts have described, in Reuters' framing, as "circular" financing, wherein companies invest in one another and simultaneously sign supply agreements, inflating apparent demand and reported revenue without necessarily generating equivalent net economic value.
Nvidia: $30 Billion for Guaranteed Demand
Nvidia's participation in this round was the subject of intense speculation for months, with reports in September 2025 suggesting a potential investment as large as $100 billion before those numbers contracted significantly as the deal took shape in early 2026. What ultimately crystallized was a $30 billion commitment paired with specific compute capacity guarantees from OpenAI:
- 3 gigawatts of dedicated inference capacity on Nvidia's Vera Rubin systems
- 2 gigawatts of dedicated training capacity on Vera Rubin systems
Nvidia CEO Jensen Huang acknowledged the fundamental circularity of the arrangement in January, confirming that Nvidia "will invest a great deal of money" in OpenAI while OpenAI CFO Sarah Friar noted in a separate context that much of what Nvidia invests will ultimately return to Nvidia through GPU purchases — a dynamic that both parties appear comfortable characterizing as mutual alignment rather than financial engineering.
The strategic logic for Nvidia is nonetheless coherent: locking in OpenAI as a committed, large-scale buyer of Vera Rubin architecture — Nvidia's next-generation platform succeeding Blackwell — is worth considerably more than the face value of the investment, because OpenAI's compute appetite represents one of the few demand signals capable of moving meaningful volume at the scale Nvidia operates and needs to justify continued chip development at the frontier.
SoftBank: $30 Billion, $64.6 Billion Total
SoftBank's cumulative investment in OpenAI has now reached $64.6 billion, representing an ownership stake of approximately 13%, with the preferred shares acquired through this round structured to convert into common shares upon an IPO — an arrangement that aligns SoftBank's financial returns directly with the timing and success of OpenAI's eventual public offering.
To fund its continued participation, SoftBank has liquidated stakes in existing investments, including portions of its Nvidia holdings, reflecting the degree to which Masayoshi Son has repositioned the firm's portfolio around what he has repeatedly characterized as the defining technological moment of the century. His approach here is structurally consistent with the Vision Fund's original thesis — identify a transformative platform shift before consensus forms, concentrate capital at the infrastructure layer with disproportionate scale, and be positioned for the compounding returns that follow market convergence. Whether that judgment proves prescient or premature will ultimately be decided at the IPO.
What OpenAI Says It Will Do with the Money
OpenAI's stated financial targets provide a striking picture of the company's internal calculus: the company projects $600 billion in total compute spend through 2030, alongside $280 billion in revenue over the same period and approximately $100 billion in cumulative losses — figures that, taken together, represent one of the most capital-intensive bets ever made by a technology company on the proposition that scale today will generate sustainable margin tomorrow.
The capital deployment breaks down across four primary categories:
- Compute infrastructure OpenAI has now committed to over $500 billion in disclosed cloud capacity across multiple providers — $250 billion with Microsoft Azure under renegotiated terms, an expanded $138 billion with AWS, and a reported $300 billion Oracle commitment beginning in 2027 — a portfolio of infrastructure agreements that collectively represent some of the largest technology procurement arrangements in history and that effectively define the physical architecture through which OpenAI's models will reach its users.
- Inference costs Inference – the compute required to generate a response each time a user submits a query – represents the company's single most significant and least controllable operating expense, and it scales directly with the usage growth that investors are simultaneously paying to accelerate. Leaked financial documents showed OpenAI spent $8.67 billion on inference alone in the first three quarters of 2025, with projections placing that figure at $14.1 billion in 2026, in a dynamic where every new user and every more capable model iteration drives costs higher even as it validates the product's commercial momentum.
- Research and model development Maintaining a credible frontier position against Anthropic, Google DeepMind, and an accelerating open-source ecosystem requires continuous investment in training runs that individually cost hundreds of millions of dollars, consume enormous compute resources, and must be repeated as new architectures and capabilities emerge — meaning that research spend is not discretionary in any meaningful sense for a company whose identity and competitive position rest on being at the frontier.
- The IPO runway The round is explicitly positioned by OpenAI as coming ahead of an expected IPO later this year, and the financial logic is straightforward: public market investors scrutinize balance sheets with a rigor that private investors rarely apply, and a company projecting $14 billion in annual losses cannot present itself credibly to institutional equity buyers without the kind of cash buffer that makes the path to profitability appear traversable rather than theoretical.
The Numbers Behind the Valuation
At $840 billion post-money, OpenAI is being valued at roughly 60 to 65 times trailing revenue — a multiple that places it in a category of its own relative to any meaningful peer comparison.
| Company | Revenue Multiple |
|---|---|
| OpenAI | ~60–65x |
| Nvidia | ~30x |
| Microsoft | ~12x |
What justifies the premium — at least in theory — is the extraordinary velocity of OpenAI's growth trajectory: the company generated $13 billion in revenue for 2025, up from $3.7 billion in 2024, representing approximately 236% year-over-year growth and a pace of scaling that surpassed the revenue ramp of Google, Facebook, and virtually every other software company for which comparable benchmarks exist, with the company crossing $12 billion in annualized recurring revenue by July 2025 — a milestone that took Google seven years to reach and Facebook six.
Key user metrics as of February 2026:
- 900 million+ weekly active users
- 50 million+ consumer subscribers
- 9 million+ paying business users on ChatGPT
- 1.6 million weekly Codex users (tripled since the start of 2026)
The Profitability Problem
The growth story is real and well-documented, but the profitability picture is equally real — and points in a direction that demands honest analysis rather than being footnoted past.
| Metric | Figure |
|---|---|
| 2026 projected loss | $14 billion |
| Cumulative losses through 2029 | $115 billion |
| Gross margins | ~40% |
| Microsoft revenue share | 20% of total revenue, owed through 2032 |
| 2026 projected inference costs | $14.1 billion |
OpenAI's gross margins are structurally constrained by variable compute costs that increase with every additional user and every increment of model capability, creating a perverse dynamic in which the company's most successful growth outcomes simultaneously drive its most significant cost pressures. The 20% Microsoft revenue share, which was renegotiated and extended through 2032 as part of October 2025 partnership restructuring, means that a material portion of every dollar earned is pre-allocated to a partner before any operating expense is addressed.
The central thesis underlying all of this spending is that compute costs will fall as the hardware market matures, inference efficiency will improve as model architectures optimize, and revenue growth will eventually compound faster than the cost base expands — but that timeline stretches across multiple years rather than quarters, and the distance between the company's current financial position and the point at which that thesis resolves is precisely where the risk lives.
The Microsoft Question
Microsoft's conspicuous absence from this round is one of the most analytically significant details in a transaction full of them, and the careful language in the joint statement both companies released suggests that the terms of their relationship required some work to characterize accurately.
- Microsoft Azure remains the exclusive cloud provider for OpenAI's stateless APIs – the standard transactional connections in which an application sends a prompt, receives a response, and the session concludes
- AWS is now the exclusive third-party provider for stateful connections — the architecture that maintains context and memory across interactions, which is the underlying infrastructure for the most capable and commercially valuable agentic workflows
That technical distinction carries substantial strategic weight, because stateful interactions represent the architecture through which complex, multi-step, context-aware agent systems operate — which is to say, the future of enterprise AI applications as the industry currently understands it. By securing exclusive third-party access to OpenAI Frontier's stateful infrastructure through AWS, Amazon has secured a structurally privileged position in the highest-value segment of OpenAI's product roadmap, while Microsoft retains its established position in the current-generation API business that powers the majority of today's integrations.
Microsoft is now watching a relationship into which it has invested over $13 billion gradually become more architecturally complex, with Amazon occupying an increasingly central position in precisely the AI application category that enterprise technology buyers care most about in 2026. Whether Microsoft's absence from this round reflects a deliberate strategic calculation or an inability to agree on terms, the outcome is that its voice in OpenAI's strategic direction has become measurably smaller.
The Competition This Capital Is Responding To
The fact that OpenAI raised $40 billion just twelve months ago — and has now returned to private markets for an additional $110 billion — is not the behavior of a company operating from a position of unchallenged dominance. It is the behavior of a company navigating intensifying competition across multiple dimensions simultaneously, each of which carries implications that the growth figures in isolation do not fully reveal.
Anthropic's Enterprise Lead
Anthropic closed its own $30 billion Series G in February 2026 at a $380 billion valuation, with annualized revenue that grew from $1 billion at the end of 2024 to more than $5 billion by August 2025, and with 2026 projections ranging from $20 to $26 billion — a trajectory that has made it the primary beneficiary of enterprise AI spending by companies looking for alternatives to OpenAI's increasingly crowded platform.
- Annualized revenue: $1 billion (end-2024) growing to $5 billion+ (August 2025)
- 2026 revenue projections: $20 to $26 billion
- Enterprise market share: systematically taking ground from OpenAI across key verticals
One analysis cited by Reuters noted that OpenAI's enterprise AI market leadership had declined from 50% to 34% as Anthropic and Google gained ground — a contraction that is strategically more painful than it might appear in isolation, because enterprise contracts represent the most defensible, highest-margin, and lowest-churn revenue base available in AI, and losing share in that segment while simultaneously burning capital at the infrastructure layer is precisely the combination that challenges long-term financial sustainability.
Google's Model Capability
Google's Gemini 3, released in November 2025, reportedly outperformed competing frontier models across 19 out of 20 industry benchmarks and surpassed GPT-5 Pro on Humanity's Last Exam — a result that, by multiple accounts, triggered an internal "code red" response at OpenAI in December, echoing with deliberate irony the moment in 2022 when Google itself declared code red in response to ChatGPT's launch.
Google's structural competitive advantage, however, is not primarily its model quality — it is the unmatched breadth of its distribution infrastructure, through which Gemini can reach every Android device, every Google Search interaction, every Gmail and Google Workspace session simultaneously, creating a deployment surface that no frontier AI lab, including OpenAI, can replicate through any combination of capital or product excellence alone.
The Open-Source Pressure
DeepSeek's R1 model, released in early 2025, delivered an empirically unsettling demonstration that highly capable reasoning models could be trained at a fraction of the cost that frontier labs had previously assumed to be necessary — and the open-source ecosystem has been accelerating on the back of that revelation ever since, with each new capable open-source release making it incrementally more difficult to sustain the commercial proposition that proprietary frontier models justify the significant price premium they carry over freely available alternatives.
The Circular Financing Problem
The circular financing concern deserves more direct treatment than it typically receives in coverage of this round, because the mechanism by which much of the $110 billion will return to the investors through supply agreements meaningfully changes how the headline number should be interpreted.
The Amazon cycle:
- Amazon invests $50 billion in OpenAI as equity
- OpenAI commits to spending $138 billion on AWS infrastructure over eight years
- Amazon records that cloud revenue on its income statement as genuine business activity
- OpenAI records the corresponding spend as operating expense, constraining its path to profitability
The Nvidia cycle:
- Nvidia invests $30 billion in OpenAI as equity
- OpenAI commits to 5 gigawatts of compute capacity on Vera Rubin systems
- OpenAI's CFO has acknowledged that much of what Nvidia invests will return to Nvidia through GPU procurement at the scale implied by those commitments
None of this represents fraud or misrepresentation — these are genuine transactions that create genuine physical infrastructure with genuine economic value. But the optical and analytical effect is that the $110 billion headline investment figure substantially overstates the net new capital available to OpenAI for discretionary deployment toward research, talent, product development, and operating expenses, because a material share of it is committed in advance to the very companies writing the checks.
The analytically more useful question is not how much money OpenAI raised in this round, but how much net free capital remains available after the company has honored its compute commitments, fulfilled its Microsoft revenue-sharing obligations, and serviced the operating costs of running a product used by 900 million people weekly — a number that has not been disclosed and that would substantially revise the picture if it were.
The IPO Runway
Taken together, the structural features of this round form a coherent and intentional narrative aimed at a public offering: the $840 billion post-money valuation establishes an S-1 reference point, Amazon's $35 billion second tranche is explicitly triggered by an IPO filing, SoftBank's preferred shares convert to common equity at the moment of public listing, and OpenAI's nonprofit foundation — whose stake in OpenAI Group increases to over $180 billion in value as a result of this round's pricing — gains the paper wealth it needs to underwrite its AI safety and philanthropic mission at a scale proportionate to its ambitions.
The structural IPO signals:
- $840 billion post-money valuation anchoring the expected S-1 pricing range
- Amazon's $35 billion tranche explicitly triggered by an IPO filing notification
- SoftBank's preferred shares converting to common stock upon public listing
- OpenAI's nonprofit foundation stake rising to $180 billion+ in paper value
Internal targets reportedly include a confidential SEC filing in H2 2026 and a public listing targeting a $1 trillion valuation in 2027 — which would make it one of the largest IPOs in U.S. market history and would create a liquid market for the shares currently held by employees, early-stage investors, and the foundation, all of whom have been sitting on paper wealth of growing magnitude without a mechanism to convert it.
Getting there credibly requires OpenAI to present public market investors with a convincing account of how a company projecting $115 billion in cumulative losses through 2029 reaches positive cash flow by 2030 — a story that depends on favorable compute cost curves, continued user growth, enterprise market stabilization, and an AI industry that continues to expand faster than the competitive field narrows margins.
The Honest Assessment
What OpenAI has going for it:
- More private capital raised than any company in the history of technology
- 900 million weekly users and 50 million+ paying subscribers representing a genuinely unprecedented consumer AI adoption curve
- Revenue growth velocity that surpassed every meaningful historical benchmark in software company scaling
- A product that has materially and demonstrably reshaped how millions of professional and creative workers approach their daily tasks
What OpenAI is also carrying:
- $14 billion in projected operating losses for 2026 alone, with $115 billion in cumulative losses expected through 2029
- Inference costs that scale structurally upward with every new user and every capability improvement
- A 20% revenue share owed to Microsoft through 2032 that pre-allocates a material portion of income before any operating expense is addressed
- A substantial portion of this round destined to cycle back to the investing parties through compute supply agreements, meaning net free capital is considerably smaller than the headline suggests
- Declining enterprise market share under sustained, well-funded competitive pressure from Anthropic and Google
- A path to profitability that extends into the 2030s and depends on compute cost curves, competitive outcomes, and market dynamics that remain genuinely uncertain
The capital buys OpenAI time, infrastructure, the ability to compete at the frontier without being outspent into irrelevance, and the IPO optionality it needs to convert its extraordinary paper valuation into the liquid capital that would allow its founders, employees, investors, and nonprofit foundation to realize the financial outcomes they have been accumulating on paper for years.
What it does not buy is certainty — because the $840 billion valuation is ultimately a thesis about a future in which OpenAI wins, in which the scaling hypothesis continues to generate meaningful capability improvements, and in which being the default AI platform for 900 million users today translates into the dominant commercial AI platform for a comparable paying user base in five years' time.
Those assumptions are neither irrational nor impossible — but the next chapter of this story will not be written in funding announcements. It will be written in IPO prospectus language, in quarterly earnings reports subject to public market scrutiny, and in the answer to the question that institutional investors will eventually have to provide to their own stakeholders: does the underlying revenue growth justify a valuation that currently exceeds the combined market capitalization of most Fortune 100 companies, or does this resolve the way previous eras of extraordinary technology capital concentration have resolved when the fundamentals finally caught up with the financing?
FAQ
How much did OpenAI raise and at what valuation?
OpenAI raised $110 billion in a round announced February 27, 2026, at a $730 billion pre-money valuation, or $840 billion post-money including the new capital. The round was led by Amazon ($50 billion), Nvidia ($30 billion), and SoftBank ($30 billion), with the round still open and additional investors expected to join as it progresses.
Is this the largest private funding round in history?
According to Crunchbase, the $110 billion raise is the largest venture funding round ever recorded, surpassing OpenAI's own prior record of $40 billion set with the SoftBank-led round that closed in March 2025 — making OpenAI the only company in history to have broken the all-time private capital formation record twice within a twelve-month period.
What are the conditions on Amazon's $50 billion investment?
Amazon has made an initial commitment of $15 billion, with the remaining $35 billion contingent on conditions that are partially redacted in public filings. An OpenAI IPO filing is one explicitly disclosed trigger that would require Amazon to deploy the remaining amount, while a second set of conditions remains undisclosed. The equity commitment carries an expiry date of December 31, 2028, after which the obligation ends if the triggers have not been activated.
Is Microsoft still invested in OpenAI?
Microsoft did not participate in this round but retains its existing equity stake from prior investments totaling over $13 billion. Under the terms of the existing partnership, Microsoft Azure remains the exclusive cloud provider for OpenAI's stateless APIs, while AWS has been designated the exclusive third-party cloud provider for OpenAI Frontier — the enterprise agent platform that represents the most strategically important segment of OpenAI's product roadmap. OpenAI committed to $250 billion in Azure spending through 2032 as part of renegotiated partnership terms in October 2025.
How much money is OpenAI actually losing?
OpenAI projects $14 billion in operating losses for 2026 alone, with cumulative losses expected to reach approximately $115 billion through 2029 before the company achieves positive cash flow, which it is currently targeting for sometime in the 2030s. Inference costs — the compute required to process user queries across a product used by 900 million people weekly — are projected at $14.1 billion in 2026, representing one of the fastest-growing line items in the company's operating structure.
What does OpenAI plan to do with $110 billion?
OpenAI is targeting $600 billion in total compute spend through 2030 to build and maintain the infrastructure it believes is necessary to remain at the frontier of AI capability at global scale. Additional capital is directed toward ongoing research and model development, enterprise product expansion including the rapidly growing Codex coding platform, and establishing the balance sheet strength required to support a credible IPO filing in 2026 or a public listing in 2027.
What is the circular financing concern?
The circular financing concern refers to the structural dynamic in which a significant portion of the $110 billion investment will flow back to the investing parties through supply agreements — specifically, Amazon's investment paired with OpenAI's $138 billion AWS commitment over eight years, and Nvidia's investment paired with OpenAI's commitment to 5 gigawatts of Vera Rubin compute capacity. The practical implication is that the net free capital available to OpenAI for discretionary research, operations, and product development is considerably smaller than the headline investment figure suggests, though the precise amount has not been publicly disclosed.
When is OpenAI's IPO expected?
Internal targets reportedly include a confidential SEC filing in H2 2026 and a public listing in 2027 targeting a valuation of up to $1 trillion. The Amazon deal structure — in which the $35 billion second tranche is explicitly triggered by an IPO filing — creates a significant and time-bounded financial incentive for OpenAI to complete the offering within the equity commitment's December 31, 2028 expiry window.
Related Articles
Dani
Mark
Dani
