OpenAI projected to bring in $20bn less in revenue than expected

OpenAI told investors that revenue this year would reach $50bn, $20bn below the $70bn figure it had signalled last month, raising concerns about AI demand growth. The gap partly reflects different accounting for cloud-partner sales: Anthropic includes them, while OpenAI does not. The news coincided with a 1.4% drop in the Nasdaq on Thursday; Nvidia fell 2.9%, Oracle 5.5% and Micron 4.8%. OpenAI is in early-stage talks to raise $30bn at a roughly $1.4tn valuation.

Analysis

The key risk is not the headline-sized revenue gap itself, but whether investors mistake a reporting-basis mismatch for weaker end demand—or, conversely, dismiss a genuine forecast reset as accounting noise. Cloud-partner sales make Anthropic’s measure look more inclusive; verify comparable definitions and cash economics before extrapolating either company’s figures to total AI demand. For AMZN and GOOG, partner-channel revenue may support reported cloud growth, but does not establish incremental profit: capacity, pricing, and the share of workloads captured matter. ORCL is more exposed to any pause in customer commitments if AI infrastructure spending slows. NVDA and MU face second-order risk through delayed accelerator and memory orders, but one private-company forecast is not evidence of industry-wide capex cuts.

Near term (days), the initial tech selloff can extend through crowded AI positioning; avoid treating it as confirmation of a demand break. Over 1–3 months, comparable revenue definitions, cloud growth, and hyperscaler capex commentary should resolve the signal. Anthropic’s possible IPO could sharpen scrutiny of forward revenue quality and private-market valuations. Over 6–18 months, safety regulation and execution risks could slow deployment, while continued enterprise adoption would support infrastructure demand. The contrarian point: this may be a measurement and fundraising-narrative shock rather than an immediate change in compute consumption. A genuine thesis break would be multiple hyperscalers cutting capex or weaker cloud growth, not this disclosure alone.

Key Decisions for Investors

• Do not chase the initial NVDA/MU/ORCL decline as a standalone short signal. Keep AI exposure sized conservatively until hyperscaler capex guidance and cloud growth provide independent confirmation.
• For an immediate hedge, reduce the most momentum-sensitive AI hardware exposure or use a modest, defined-risk semiconductor hedge; remove it if major cloud providers maintain or raise capex plans and demand indicators remain firm.
• Treat AMZN and GOOG as potential relative beneficiaries only if cloud disclosures show monetized AI workloads and attractive economics. Do not infer earnings upside solely from their role as distribution partners.
• Set a 1–3 month alert for Anthropic IPO disclosures and comparable revenue definitions. Reassess private-market valuation risk if reported growth relies materially on partner sales without corresponding evidence of cash conversion or durable customer demand.