OpenAI closes $122B round — largest private financing in history
On March 31, 2026, OpenAI officially closed the mega-round that had been running since February — at a post-money valuation of $852 billion and a total volume of $122B. That makes it the largest private financing round ever completed in the tech industry — about ten billion more than initially announced in February. SoftBank co-led with Amazon, Nvidia, Andreessen Horowitz, and D. E. Shaw Ventures.
Round at a glance
- Size: $122B (originally announced at $110B, revised up to $122B in March).
- Valuation: $852B post-money, $730B pre-money.
- Lead investors: Amazon ($50B, partly contingent on IPO/AGI triggers), Nvidia ($30B), SoftBank ($30B).
- Revenue base: about $2B monthly run rate, $13.1B for fiscal year 2025.
- Close: March 31, 2026.
Where things stood
OpenAI had run several large rounds across 2024–2025 — most recently at around $300B valuation. In early 2026 a new tier became visible: Microsoft, the dominant strategic partner until then, stepped back from the lead position, and a broader investor consortium took shape. February 2026 brought the announcement: $110B round at $730B pre-money — already among the largest private financings ever.
The final size was not yet locked. Reports talked about $110B with an option to increase. Lead investor was Amazon with a $50B commitment, followed by Nvidia and SoftBank with $30B each. Microsoft, for years the biggest single backer at ~$13B, did not feature in this constellation.
What stands now
1. $122B committed, $852B valuation. On March 31 the round was officially closed — $12B more than reported in February. Bloomberg and CNBC independently confirmed: final size sits at $122B. SoftBank co-led, with Andreessen Horowitz and D. E. Shaw Ventures among further co-investors. OpenAI framed this in its own statement as “a new phase of AI.”
2. Amazon’s $50B — mostly tied to triggers. Of Amazon’s $50B commitment, only about $15B is drawn immediately, per Bloomberg reporting. $35B are contingent on two possible triggers: IPO or reaching an AGI definition under OpenAI’s structuring agreement with Microsoft. This structure matters because the effective cash line in the short run is much smaller than the headline number.
3. SoftBank and Nvidia at $30B each — strategic threads. Nvidia is not just contributing capital — it is supplying hardware capacity. The investment is part of a broader package securing OpenAI’s compute infrastructure over the next several years. SoftBank’s $30B flow through the Vision Fund apparatus, with the long payout choreography typical of SoftBank deals.
What is barely discussed
Public discussion is dominated by the $122B headline. The structural detail that a substantial portion only flows on certain triggers is often glossed over. That matters because the actual liquidity OpenAI can deploy over the next 12–18 months is meaningfully lower than the headline volume. Reuters reporting suggests only about $40–50B of the $122B is drawn immediately — the rest spreads across tranches or comes with conditions attached.
Second observation: Microsoft’s absence as the largest single investor. The years-long “Microsoft–OpenAI axis” softens — structurally it remains via Azure compute contracts and the Microsoft stake, but in the ownership mechanics of the latest round Microsoft loses weight. From a geopolitical lens: with SoftBank, Amazon, and Nvidia, the constellation is now far more US/Asia balanced than before.
What stands — outlook
With this round, OpenAI mainly funds compute — Stargate datacenters, inference infrastructure, chip contracts with Nvidia and AMD. The money is not primarily for research but for scaling delivery. Concretely: the next 12–24 months will be shaped less by the next model release than by buildout of backbone infrastructure.
From a user perspective the most important consequence is not the valuation but what it forces: OpenAI must monetize the capacity now being financed. Expect further tier differentiation, enterprise bundles, possibly new API pricing structures over the next quarters — comparable to what we already saw at Anthropic across 2025/2026.
What you can do now
If your stack rides on OpenAI: expect further pricing and capacity adjustments through 2026. An $852B valuation forces margin discipline — and that plays out in tier structures, rate limits, and premium features.
If you are planning multi-model setups: this market block — OpenAI, Anthropic, Google — will continue at this investment pace over the next 12 months. Other vendors (Mistral, DeepSeek, local open-weight models) remain relevant, particularly for cost-sensitive workloads that cannot absorb 20–30% pricing volatility.
If you advise as an agency: the valuation dynamic is a signal that tooling lock-in becomes a strategic risk. Building workflows so that model swaps don’t require fundamental rebuilds is no longer a theoretical concern in 2026.
Entdecke mehr
Three Days with Claude Fable 5 — and Why Opus Stays My Daily Driver
Fable 5 fixed a bug in one shot that Opus failed at twice. I'm still not switching. My honest impressions from the trenches after three days.
GlossarGPT-4o
GPT-4o (May 2024) was OpenAI's first natively multimodal model — text, audio and image in a single model, with a real-time voice mode and substantially lower cost than GPT-4 Turbo.
LexikonComparing AI models — who builds what and how to choose
The major model families in 2026 at a glance. Who builds Claude, GPT, Gemini, Llama, Mistral, DeepSeek, Qwen — and which model to pick when.