OpenAI Valuation: From $300B to a $730–850B IPO Filing

OpenAI filed confidentially for an IPO on June 8, 2026 — Goldman Sachs and Morgan Stanley leading — targeting a $730–850 billion valuation. On August 10, 2026 the company closed a $7 billion employee tender offer that reaffirmed its $852 billion price tag, while CFO Sarah Friar has since told associates OpenAI now leans toward a 2027 listing rather than the September 2026 debut once floated. The $300 billion price from its early-2025 secondary tender, a record at the time, is now just a waypoint on the chart.

Annualized revenue topped $40 billion in July 2026 according to Bloomberg (up from ~$20B at the end of 2025, ~$6B in 2024, and ~$2B in 2023), and yet Anthropic's revenue run rate is still ahead at roughly $65B by the end of July 2026, on a $965B valuation set in its June 2026 IPO filing. The rest of this post walks through how the $300B era was priced, what changed by mid-2026, and what the August 2026 tender and the 2027 IPO lean mean now.

OpenAI Valuation 2026: The 60-Second Breakdown

OpenAI was valued at $300 billion in its October 2025 secondary tender led by SoftBank, with participation from Thrive Capital, Dragoneer Investment Group, and Abu Dhabi's MGX. That price reflected roughly 23x its then-$13B annualized revenue run-rate, a doubling from the $157B October 2024 round. By March 2026 the company had repriced far higher — to $852 billion via a $122B primary round — and filed confidentially for an IPO on June 8, 2026 targeting $730–850B. On August 10, 2026, a $7 billion employee tender offer reaffirmed the $852B price unchanged, even as the IPO timeline itself has slipped toward 2027. The breakdown below explains how that $300B mark was justified and what carried it forward.

How the $300B OpenAI Valuation Was Set

The $300B figure is not a primary-round price — it's the per-share value of a structured secondary tender that let employees and early investors sell shares to SoftBank-led buyers. Tender offers usually clear at a discount to the headline number, but in OpenAI's case the tender priced at the same per-share level as the primary, which is itself unusual.

The valuation arc since 2019 tells the story better than any single number:

That's roughly a 947x increase in under seven years, from the $0.9B capped-LLC structure to the $852B mark the August 2026 tender reaffirmed. Even the AI-era curve from $29B (April 2023, two months after GPT-4) to $852B (March 2026) is about 29x in 35 months.

Revenue Mix: From $13B to $40B+ ARR

When critics call OpenAI overvalued they usually focus on the multiple. But the underlying revenue mix has changed dramatically — and it's no longer just ChatGPT Plus subscriptions. The breakdown below is the last fully itemized run-rate snapshot (~$13B); the total has since roughly tripled to over $40B annualized by July 2026, per Bloomberg, with Codex and ChatGPT Work driving a growing share of the API and enterprise lines:

Consumer revenue (Plus + Pro + Team) is now ~74% of total — that's the highest consumer mix of any frontier AI company by a wide margin. More detail on revenue trajectory here.

OpenAI Valuation 2026 vs Anthropic, xAI, and the Frontier Peers

As of August 2026, OpenAI's revenue multiple is no longer the lowest in the cohort — Anthropic's faster revenue growth has pushed its multiple below OpenAI's even at a higher headline valuation. Compare like-for-like:

Figures blended from Bloomberg reporting on OpenAI's July 2026 run rate and August 2026 tender, Reuters/Axios reporting on Anthropic's end-of-July run rate, and Sacra/company disclosures for the smaller peers; multiples are approximate, calculated on the latest disclosed valuation against the most recent reported annualized revenue. ~21x is still reasonable next to public comps — Palantir at ~70x sales, CoreWeave at ~30x — but it's no longer the cheapest seat at the frontier table now that Anthropic's revenue has scaled faster. See the full peer breakdown on the AI Valuations dashboard.

What Justifies the OpenAI 2026 Valuation

The bull case rests on four specific data points, each of which justifies meaningful multiple expansion on its own:

The Bear Case Against the $300B OpenAI Valuation

The bear case is concrete and worth taking seriously — most $300B price tags collapse under one or two of these risks, and OpenAI has all four:

August 2026: The $7B Tender, and Why the IPO Now Leans 2027

OpenAI confidentially submitted its IPO prospectus to the SEC on June 8, 2026, without disclosing a listing date. Weeks later, the New York Times reported that the company was weighing two paths: list in 2026 at a lower valuation, or wait until 2027 for a shot at $1 trillion. CEO Sam Altman reportedly called any cut below $1 trillion a "nonstarter," and CFO Sarah Friar has since told associates OpenAI is now targeting a 2027 debut — a real change from the September 2026 window once discussed. On August 10, 2026, OpenAI closed a $7 billion tender offer, buying back employee and former-employee shares itself (rather than bringing in outside investors) at a price that reaffirmed the $852 billion valuation set by its March 2026 primary round, unchanged three-and-a-half months later. It was OpenAI's third structured liquidity event for employees, after a $1.5B tender in 2024 and a $6.6B tender in October 2025.

The sequence now looks like this:

A confidential S-1 is a step toward an IPO, not a commitment — OpenAI can still delay or withdraw it depending on market conditions. For retail investors wondering how to get exposure before then, options remain limited: the SpaceX pre-IPO playbook applies almost identically here — secondary marketplaces like Forge Global (where shares traded around $721.85 as of August 20, 2026) and EquityZen, accredited-only SPVs, and indirect exposure through Microsoft and SoftBank stock.

What the $300B OpenAI Valuation Means for VCs and Founders

Three implications I think matter most for anyone investing in or building AI right now:

1. The frontier AI moat is now distribution, not model quality. OpenAI is the most-valuable AI company because it has 700M users, not because GPT-5 is meaningfully better than Claude 4 or Gemini 2.5. Founders building AI startups should optimize for the user, not the benchmark.
2. Roughly 15-21x ARR is the frontier ceiling at $500B+ scale, not the floor. Both OpenAI (~21x on $852B/$40B) and Anthropic (~15x on $965B/$65B) now sit in that band. AI startups doing $10M ARR at 100x multiples are repricing toward 30x as they grow — meaning growth has to outrun multiple compression. The same dynamic shows up well outside pure software: Genesis AI's reported $3B valuation talks price the robot-foundation-model startup at roughly 29x its year-old seed round, almost entirely on narrative rather than revenue.
3. Capex commitments are the new dilution — track the numbers on the AI Spending dashboard. OpenAI committed $500B+ in compute against a $40B+ run rate as of July 2026 — still roughly 12x revenue. The capital intensity of frontier AI means even winners give up enormous equity — Microsoft has ~49%, SoftBank growing. That dynamic doesn't exist in SaaS.