OpenAI annualised revenues $20B less than previously signalled
OpenAI’s previously touted $70 billion “annualized” revenue figure appears to be roughly $20 billion too high, after new investor materials reportedly showed closer to $50 billion and revealed differing accounting treatments versus rival Anthropic. Commenters argue that such annualized run-rate numbers are inherently flimsy and potentially misleading, especially when used to justify near–trillion‑dollar valuations ahead of anticipated IPOs. The exchange widens into concerns about opaque private-market financials, the sustainability of current AI business models and capex, and whether the broader AI boom resembles a speculative bubble that could end badly for investors.
Revenue discrepancy & “annualized” metrics
- Thread focuses on reports that OpenAI’s annualized revenue was previously signaled ~$70B but investor materials show closer to ~$30B in July.
- Posters debate whether this is outright deception, investor “gross-up” for comparability with Anthropic, or media misinterpretation of vague investor guidance.
- Many criticize “annualized revenue” based on peak days/weeks as inherently misleading; several call it “bullshit revenue” used to hype valuations.
- Some argue this could border on securities fraud if used to raise money; others say as a private firm using fuzzy language with press, it’s legally safer.
Comparison with Anthropic & accounting games
- Anthropic’s figures reportedly include cloud-partner revenue (AWS, Google) and sometimes exclude big training costs, which commenters call “profit before bad things.”
- OpenAI reportedly does not usually include cloud-partner revenue, so attempts to “gross up” for comparison produced inflated numbers.
- Many see both labs as using non‑GAAP, selectively constructed metrics to appear profitable or hyper‑growth.
Business model viability & AI bubble concerns
- Strong skepticism that current LLM businesses justify near‑trillion or multi‑trillion valuations, given huge training/inference costs and relatively modest paying user base.
- Some report very high API spend in their own companies; others note usage is often justified by persuasion, not hard ROI metrics.
- Several predict a major AI investment bust akin to dot‑com (but not 2008‑style systemic), with wealthy investors taking most of the hit.
- Others counter there is at least a solid, if smaller, cloud AI utility business that will survive even if valuations collapse.
IPO timing, private markets & regulation
- Many argue that if OpenAI’s fundamentals were strong and stable, it would have already IPO’d; delaying is seen as a red flag.
- Others note many successful large companies stay private (helped by regulatory changes and abundant late‑stage private capital), so non‑IPO ≠ weak business.
- Discussion covers changes like higher shareholder thresholds and JOBS Act effects that let huge firms stay private with limited disclosure.
AI, disinformation & public narrative
- Some highlight LLMs as “bullshit/disinformation machines”: persuasive but error‑prone outputs and use in propaganda campaigns.
- Others push back, saying that characterization is exaggerated or reflects misunderstanding of hallucinations.
- Several criticize both labs and media for sophisticated hype cycles, anonymous sourcing, and selective leaks that shape perception ahead of IPOs and funding rounds.