OpenAI $122B Round at $852B Valuation: Amazon, Nvidia, SoftBank; IPO Path to $1T

OpenAI closed the largest private funding round in history in late March 2026: $122 billion at an $852 billion post-money valuation. The lead investors are Amazon (a large contingent commitment, portions tied to IPO or AGI milestones), Nvidia and SoftBank (each in the tens of billions), plus sovereign wealth funds, institutional players, and approximately $3 billion from retail investors via platforms including Republic and Forge. OpenAI has topped $25 billion in annualized revenue — reached in 39 months, faster than any software company in history. CFO Sarah Friar is targeting a 2027 IPO listing, with some advisers believing a late 2026 SEC filing is possible. The for-profit conversion, required to enable a public listing, is already completed.

OpenAI crossed $25 billion in annualized revenue in roughly 39 months from its late 2022 baseline. Google took 17 years to reach $25 billion in annual revenue. Salesforce took 18 years. Facebook took 12. No software company has scaled to this revenue level this quickly. The context makes the $852 billion valuation — approximately 34x annualized revenue — more legible as a forward multiple on a hypergrowth trajectory rather than a stretch on current fundamentals.

The $122 Billion Round: Investor Breakdown

The round structure reflects the competitive dynamics across the AI infrastructure stack:

Amazon committed a large contingent sum with portions tied to IPO completion or AGI milestone achievement by December 2028. Amazon's commitment partially overlaps with its separate $25 billion Anthropic deal — Amazon is simultaneously the largest investor in both OpenAI and OpenAI's primary competitor, hedging across the frontier model stack.

Nvidia invested approximately $30 billion. For Nvidia, this is both a financial return play and a strategic alignment: OpenAI's GPT-6 training runs will consume enormous quantities of Nvidia hardware, and an Nvidia equity stake in OpenAI creates alignment on roadmap timing and hardware allocation.

SoftBank invested approximately $30 billion. SoftBank's Masayoshi Son has positioned this as part of a broader "Project Izanagi" AI infrastructure bet — SoftBank is investing in OpenAI, contributing to the Stargate data centre programme, and separately funding AI infrastructure across Asia.

Retail investors via Republic and Forge contributed approximately $3 billion — a notable first for OpenAI and a deliberate move ahead of IPO to broaden the shareholder base and demonstrate retail demand depth before a public listing.

Sovereign wealth funds: specific funds were not fully disclosed but multiple Gulf and Asian sovereign funds participated, reflecting the same pattern of state-level AI infrastructure investment seen in the Anthropic and Google TPU deals.

$25 Billion ARR: The Revenue Scaling Story

OpenAI crossed $25 billion in annualized revenue at the end of February 2026, up from $21.4 billion at end-2025 — a 17% increase in approximately two months. The trajectory implies a run rate approaching $30 billion by end-2026 if growth continues at current pace.

The revenue breakdown is not fully public, but known components include:

• ChatGPT subscriptions (Plus at $20/month, Pro at $200/month, Team/Enterprise tiers)
• OpenAI API usage across GPT-4, GPT-4o, o1, o3, o4, and GPT-5 tiers
• Enterprise direct contracts (reported multi-year deals with major financial institutions, consulting firms, and enterprise software companies)
• Operator tier partnerships (companies embedding ChatGPT into their products)

The $25 billion figure is annualized run rate, not recognised annual revenue. OpenAI recognised $13.1 billion in revenue in calendar 2025 while spending approximately $22 billion — a gross margin of 33% that reflects the cost of running frontier model inference at scale. The gap between revenue and spending is why the $122 billion raise is necessary: OpenAI is not yet generating the cash flows to fund its own compute buildout.

The IPO Path: $600 Billion Compute Target

OpenAI is targeting approximately $600 billion in total compute spending through 2030 as the capital plan that underpins GPT-6-scale training. This figure — which involves Stargate data centres, Nvidia hardware, and potentially Microsoft and Oracle cloud infrastructure — defines the investment requirement for maintaining frontier model leadership.

CFO Sarah Friar has communicated a 2027 IPO target to associates, while some advisers put the filing date as early as late 2026 depending on SEC processing timelines. The for-profit conversion — moving OpenAI from its unusual nonprofit parent/capped-profit subsidiary structure to a more conventional corporate structure — has been completed, removing the main structural barrier to a public listing.

A $1 trillion post-IPO valuation would require approximately $130-150 billion in annual revenue at current tech company revenue multiples — or a premium multiple justified by the GPT-6/AGI narrative. At current growth rates ($25 billion ARR in early 2026), $130 billion in revenue is a 2030-2031 prospect. The $1 trillion IPO scenario is not a current revenue multiple story; it is a "we are on a path to AGI and priced accordingly" story that requires investor willingness to pay a narrative premium.

What This Means for OpenAI's Competitive Position

The $122 billion round gives OpenAI approximately 4-5 years of runway at current burn rates ($22 billion per year in spending) even before API revenue growth. More importantly, it provides the capital to fund GPT-6 training runs that will require infrastructure beyond what OpenAI's prior capitalisation could support.

GPT-6 scale training — estimated at 10-100x the compute of GPT-5 — requires data centre infrastructure that does not yet exist at the required density. The Stargate programme (OpenAI, SoftBank, Microsoft, and Oracle collaborating on US-based data centre construction) is the physical infrastructure underpinning the GPT-6 training plan. The $600 billion compute spend target through 2030 is the funding requirement for that infrastructure.

For developers building on the OpenAI API stack, the $122 billion round signals three things: continued model investment (GPT-6 is funded), API pricing stability (no distress pricing pressure), and an IPO event horizon that will create new equity liquidity for OpenAI employees and early investors — which historically brings talent retention and recruitment advantages.

The Amazon Paradox: Investing in Both OpenAI and Anthropic

Amazon's position across both the OpenAI ($50 billion contingent) and Anthropic ($33 billion total) deals is deliberately constructed. Amazon's AWS infrastructure business benefits from both companies' cloud spend regardless of which frontier model wins commercial adoption. Amazon Bedrock needs both Claude and GPT-5-class models to offer enterprise customers model choice.

The contingent structure on both deals aligns the commitments with outcomes rather than creating fixed spend regardless of performance. Amazon is, effectively, buying infrastructure demand from both leading frontier model labs while paying most of the headline figures only if specific milestones (IPO, AGI, commercial thresholds) are met.

Developer Infrastructure Implications

API capacity: $122 billion in capital means OpenAI will continue expanding API infrastructure through 2027-2028. Developer platform capacity constraints — the API rate limits and availability issues that have periodically frustrated high-volume users — will be less binding as Stargate data centres come online.

Model pricing: OpenAI is not profitable at current scale. The path to profitability runs through either revenue growth outpacing compute cost growth (requires continued adoption acceleration) or compute cost reduction (requires Nvidia price drops or custom silicon). Developer pricing on the API is unlikely to rise materially before IPO — OpenAI has competitive pressure from Anthropic and DeepSeek V4 Pro open-source alternatives.

IPO event: An OpenAI IPO in 2026-2027 creates liquidity for early employees and angels. Historically, major AI lab IPOs (or near-IPOs) produce a wave of founders leaving to start new companies with liquid equity. The OpenAI IPO event horizon should be watched as a signal for the next wave of AI startup formation.

Key Takeaways

• $122 billion round closed March 2026: largest private funding round in history; $852B post-money valuation; Amazon, Nvidia, SoftBank lead investors; ~$3B retail via Republic/Forge
• $25B annualized revenue: reached in 39 months — faster than any software company in history; up 17% from $21.4B at end-2025; targeting ~$30B by end-2026
• IPO targeting 2027, possible late 2026 filing: for-profit conversion completed; CFO Sarah Friar guiding 2027; $1T valuation scenario requires narrative premium, not current revenue multiple
• $600B compute plan through 2030: Stargate data centres; GPT-6 training infrastructure; the capital plan that makes the round size legible
• Amazon invested in both OpenAI and Anthropic: contingent structures on both deals; AWS infrastructure business wins from both companies' cloud spend regardless of model winner
• Profitability gap: $13.1B revenue in 2025 vs ~$22B spend; 33% gross margin; not yet self-funding — the round bridges to the scale at which compute costs normalise relative to revenue

For the competing frontier model context, read DeepSeek V4 Pro: 1.6T Parameters, Beats Claude on Coding. For the Google-Anthropic $40B deal, read Google Invests $40B in Anthropic. For the Amazon-Anthropic deal, read Amazon $25B Anthropic: $100B AWS, Trainium Compute.