OpenAI is a San Francisco-based artificial-intelligence research and product company, the maker of ChatGPT, the GPT model family, the developer API and the Codex coding agent. Founded in 2015 as a nonprofit, it now operates as a public benefit corporation, OpenAI Group PBC, controlled by the nonprofit OpenAI Foundation. It is privately held; its shares are not listed.
expand
OpenAI launched ChatGPT in late 2022 and became one of the fastest-scaling commercial platforms ever, reaching run-rateAn annualized figure implied by extrapolating a recent shorter-period result, here OpenAI’s stated monthly revenue scaled to a year. revenue of roughly $2B per month by early 2026 (OpenAI-stated). Sam Altman is Chief Executive Officer; Bret Taylor chairs the board. The business spans four reinforcing surfaces – consumer (ChatGPT), enterprise, developers (API and Codex) and the compute that powers them – which the company describes as a single flywheel. Reported headcount was around 9,300 as of May 2026 (Tracxn, point-in-time). As a private company, OpenAI publishes selected operating figures in its own announcements but files no audited public financial statements.
The crux: can OpenAI convert an enormous valuation, scale and compute build-out into a profitable, defensible business before the capital and the competition catch up?
expand
OpenAI states ChatGPT has roughly 6x the monthly web visits and mobile sessions of the next-largest AI app, with total time spent about 4x the next app and 4x all others combined. That consumer reach is the funnel into paid and workplace use.
expand
OpenAI says it is growing revenue roughly four times faster than the platform companies that defined the internet and mobile eras. Enterprise now exceeds 40% of revenue and the API processes more than 15 billion tokens per minute.
expand
Backers span strategic partners and large institutions; OpenAI also opened the round to individuals through bank channels (over $3B) and ETF inclusion. Its infrastructure portfolio spans clouds (Microsoft, Oracle, AWS, CoreWeave, Google Cloud) and silicon (NVIDIA, AMD, AWS Trainium, Cerebras, and an own chip with Broadcom).
expand
The October 2025 recapitalization converted the for-profit into OpenAI Group PBC, removed prior fundraising limits, and aligned all equity holders on the same traditional stock. Microsoft holds about 27%; employees and other investors hold the remaining ~47%.
expand
Sacra estimates 2026 burn near $27B rising toward $63B in 2027, with inference costs alone projected around $14B in 2026 and a gross margin near 33%. Separate leaked-financials reporting put the 2025 operating loss around $21B. OpenAI itself states it is still burning cash and not profitable.
expand
OpenAI has officially committed to an incremental $250B of Azure purchases and a broad multi-cloud, multi-chip and data-center build-out (including Stargate with Oracle and SoftBank). Third-party analysis estimates total contractual outflows of several hundred billion over coming years against far smaller available liquidity – the gap an IPO is reportedly designed to help close.
expand
Anthropic (Claude) is reported around a $350B valuation with a B2B focus; Google fields Gemini on custom TPUs; Meta pursues an open-weight Llama approach; xAI and DeepSeek add further pressure. Frontier capability is increasingly converging even where capital is not.
expand
Some valuation specialists note that headline rounds blend cash with compute credits and conditional tranches, and that a multiple of this order embeds growth few companies have sustained at scale. A portion of one anchor commitment is reported to be contingent on an IPO or reaching AGI.
Reference points only, not a valuation. With no public shares, the anchors are the private funding ladder and the post-recapitalization ownership split.
expand
Because the company is private, these are not market-clearing prices. The Foundation also holds a warrant for additional equity if OpenAI’s value rises more than tenfold over 15 years. Reported IPO preparation in mid-2026 points to a $1T-plus target, with timing dependent on the PBC conversion, regulatory steps and market conditions; that target is third-party reporting and not a set price.
reported
next up
HIGH
expand
structural
risk
HIGH
expand
cadence
watch
MED
expand
contractual
watch
MED
expand
OpenAI files no audited public statements, so an income statement and free-cash-flow line are not available. The table below shows capital events and the economics OpenAI or named third parties have reported, all unaudited.
| Capital event / reported economics | Date | Figure | Source basis |
|---|---|---|---|
| Recapitalization to OpenAI Group PBC | Oct 2025 | ~$500B val. | OpenAI |
| Latest funding round (committed) | Mar 2026 | $122B | OpenAI |
| Post-money valuation, latest round | Mar 2026 | $852B | OpenAI |
| Revolving credit facility (undrawn) | Mar 2026 | ~$4.7B | OpenAI |
| Revenue run-rate | Q1 2026 | ~$2B / mo | OpenAI |
| Full-year 2025 revenue | FY2025 | ~$13.1B | OpenAI |
| Estimated cash burn¹ | 2026E | ~$27B | Sacra (est.) |
expand
OpenAI states it is still burning cash and is not yet profitable. Third-party estimates put the 2025 operating loss near $21B and 2026 cash burn around $27B (rising toward $63B in 2027), with a gross margin near 33% pressured by inference costs, and cash-flow breakeven not before about 2030. On the structure side, the OpenAI Foundation holds a 26% stake (~$130B) and controls the PBC; Microsoft holds ~27% (~$135B as-converted); employees and other investors hold the remaining ~47%. OpenAI has officially committed to purchase an incremental $250B of Azure services, and reported aggregate compute commitments run far higher. These figures are unaudited and, where attributed to third parties, are estimates that an eventual IPO prospectus would replace with disclosed numbers.
OpenAI monetizes across four surfaces: consumer (ChatGPT subscriptions, with an early ads pilot), enterprise (ChatGPT Business and Enterprise), developers (the API and Codex), and the compute that powers them. Enterprise is more than 40% of revenue and, per the company, on track to reach parity with consumer by the end of 2026.
expand
The corporate structure is a nonprofit-controlled public benefit corporation: the OpenAI Foundation holds 26% of OpenAI Group PBC and appoints its board, keeping the AGI-benefits mission in governance while allowing conventional equity and fundraising. Revenue economics are unusual for the sector: a self-reported ~$2B monthly run-rate against heavy reinvestment in compute, with a reported gross margin near 33% constrained by inference cost. OpenAI describes the surfaces as one flywheel – consumer reach funnels into enterprise, developer usage expands the platform, and compute lowers the cost per unit of intelligence over time, which the company argues should drive operating leverage as the platform matures.
OpenAI is the largest frontier-AI company by reported revenue, valuation and consumer reach, but the field is well capitalized. Its closest peers are Anthropic (Claude), Google DeepMind (Gemini), Meta (Llama) and xAI (Grok), with DeepSeek and others adding pressure.
expand
Anthropic is reported around a $350B valuation and positions Claude for enterprise and developer use; Google fields Gemini on its own TPU infrastructure and broad distribution; Meta pursues an open-weight Llama strategy; xAI and several open and regional labs compete on cost and capability. OpenAI’s edge is distribution – 900M-plus weekly ChatGPT users and a consumer brand – plus a deep capital and compute base. The competitive question is whether that distribution and capital advantage offsets a narrowing gap in raw model capability, since several labs now ship frontier-class models and capability increasingly converges even where spending diverges.
The strategy is a compute-to-product flywheel: more compute trains more capable models (the GPT-5 series, with GPT-5.4 and GPT-5.5 shipped), better models power better products and agents, and adoption drives revenue that funds the next turn. OpenAI is consolidating its surfaces into a single AI superapp.
expand
Growth levers include frontier model cadence, the Codex coding agent (2M-plus weekly users), agentic and multimodal capabilities, expansion into health, science and commerce, and an early ads pilot that the company says passed $100M in annualized revenue within weeks. The superapp would unify ChatGPT, Codex, browsing and agents into one agent-first experience, framed as a distribution strategy that turns consumer familiarity into enterprise adoption. Underpinning all of it is a diversified infrastructure portfolio – clouds across Microsoft, Oracle, AWS, CoreWeave and Google Cloud; silicon across NVIDIA, AMD, AWS Trainium, Cerebras and an own chip co-designed with Broadcom; and data-center partnerships including Stargate with Oracle and SoftBank.
OpenAI’s moat rests on distribution, model capability, compute access and ecosystem. The assessment below reads each pillar for durability and for what could erode it.
expand
Consumer distribution and brand (durable, watch). 900M-plus weekly ChatGPT users and a household-name brand are hard to replicate and feed enterprise adoption. Erosion risk: a rival app or platform default (search, mobile OS, productivity suites) capturing attention, or user habits proving less sticky than assumed.
Model capability frontier (moderate). Leading models attract usage and talent. Erosion risk: capability convergence as Anthropic, Google, Meta and xAI ship comparable models, commoditizing raw intelligence.
Compute access and scale (durable, costly). Deep capital and multi-cloud, multi-chip supply secure the compute frontier. Erosion risk: the same commitments are the source of the cash burn and funding gap; reliance on external capital and providers is a vulnerability.
Developer and enterprise ecosystem (moderate). The API, Codex and growing enterprise deployments create switching costs. Erosion risk: enterprises deliberately multi-source models to avoid single-vendor dependence, capping share.
expand
Management frames consumer adoption, enterprise deployment, developer usage and compute as one self-reinforcing system: more compute yields more intelligent models, which make better products, which drive adoption, revenue and cash flow, which fund the next turn. The company argues durable access to compute is the strategic advantage that compounds across research, products and cost of delivery.
expand
OpenAI says it is building a single agent-first product that brings together ChatGPT, Codex, browsing and broader agentic capabilities, so advances in model capability translate directly into engagement. It describes this as a distribution and deployment strategy, with consumer familiarity acting as the front door for enterprise usage.
expand
Enterprise already exceeds 40% of revenue, and the company states it is on track to reach parity with consumer by the end of 2026, with GPT-5.4 driving record engagement across agentic workflows and Codex growing rapidly among developers. This is a forward statement, not yet a reported outcome.
expand
A self-reported $2B per month annualizes to about $24B (derived), up from a $1B-per-quarter pace at the end of 2024 and ~$13.1B for full-year 2025. This is a run-rate extrapolation of an unaudited figure, shown for scale, not a projection of full-year results.
expand
Dividing the $852B post-money round value by the ~$24B annualized run-rate gives roughly 35x (derived). The figure is sensitive to both inputs – a private mark and a self-reported run-rate – and so should be read as context, not a market price; an IPO would replace it with disclosed, audited numbers.
expand
The nonprofit OpenAI Foundation holds 26% of the PBC, valued near $130B at the current round, plus a warrant for more equity if value rises more than tenfold over 15 years. The Foundation has announced an initial $25B commitment toward health and AI-resilience work, so commercial success directly funds the stated mission.
expand
OpenAI confirms it is not yet profitable. Third parties estimate 2026 cash burn near $27B, rising toward $63B in 2027, with breakeven not before roughly 2030 and a gross margin near 33% pressured by inference cost. These are unaudited estimates, but the direction – large, sustained losses funding the build-out – is not disputed by the company.
expand
OpenAI has officially committed to $250B of incremental Azure purchases and a broad data-center and chip build-out; third-party analysis estimates total take-or-pay commitments in the hundreds of billions, far above current liquidity. Closing that gap depends on revenue scaling, further capital raises and, reportedly, an IPO. If any of those slip, the commitments become a strain rather than an advantage.
expand
Anthropic, Google DeepMind, Meta and xAI all ship frontier-class models, and capability is converging even as spending diverges. Enterprises increasingly multi-source models to avoid single-vendor dependence. OpenAI’s distribution lead helps, but raw model superiority is harder to sustain as the field matures.
expand
The nonprofit-controlled PBC structure is unusual, and a public listing would test how Foundation control, mission obligations and shareholder interests coexist. Reported IPO preparation adds execution and disclosure risk: audited figures could differ from self-reported numbers, and timing depends on the conversion, regulatory steps and market conditions.
OpenAI enters mid-2026 as the largest and most-watched private company in AI, with consumer scale, a fast-rising revenue run-rate and the deepest capital and compute base in the field – set against losses measured in the tens of billions, compute commitments larger still, a thickening competitor set and an unusual governance structure heading toward a possible public listing. Whether the flywheel turns into durable profit, or the funding gap turns into a strain, is the question the next funding, model and disclosure milestones will answer.
expand
expand
expand
expand
expand
expand
expand
expand
expand
expand
expand
NOT FOR REDISTRIBUTION