Our Take
The IPO signal is real; the $34B burn rate exposes the math problem OpenAI must solve before going public: revenue growth has not kept pace with spending.
Why it matters
Investors will scrutinize unit economics and path to profitability. OpenAI's scale-first strategy (heavy capex on compute, talent, and training) works only if revenue compounds faster than costs. The company is betting it can.
Do this week
Enterprise buyers: lock multi-year API contracts now before pricing models shift post-IPO or spending pressures force margin compression.
OpenAI's $34B spending bill
OpenAI's annual operating expenses reached $34 billion in the most recent fiscal year, according to Financial Times reporting. The figure includes costs across model training, infrastructure, staff, and operations as the company prepares for an initial public offering.
The company has not disclosed revenue figures in public filings, so the ratio of spending to income remains unknown. The planned IPO will force disclosure of both metrics, making this $34B figure a stark early signal of the company's cash consumption rate.
The profitability question haunts the IPO plan
A $34 billion annual burn rate is sustainable only if revenue is tracking meaningfully higher. For context, that spending level exceeds the annual revenue of most enterprise software companies. OpenAI's business model (API access to GPT-4, ChatGPT Plus subscriptions, and enterprise contracts) has grown, but the company has not published figures to show spending growth has slowed or that unit economics are improving.
Public markets will demand answers on three fronts: (1) What is annual revenue today? (2) What is the path to operating profit? (3) How much capex is required in 2025 and beyond to maintain model quality?
The IPO timing also matters. Regulation uncertainty, customer concentration risk (estimates suggest Microsoft contracts account for a significant portion of revenue), and competition from other frontier labs (Anthropic, Google DeepMind, xAI) create headwinds. Going public while the business model is still scaling, not mature, raises execution risk.
Lock in current pricing while you can
If you use OpenAI APIs at production scale, negotiate multi-year contracts now. Post-IPO pressure to improve margins typically flows downstream to customers through price increases, usage tiering changes, or feature restrictions on cheaper tiers. Enterprise agreements signed before the IPO filing are more likely to include favorable terms and stability guarantees.
For teams evaluating alternatives, the $34B figure makes the case for multi-model strategies stronger. Build integrations with Claude, Gemini, and open-source alternatives (Llama) so you are not locked into a single vendor facing its own profitability demands.