OpenAI's London staff receive average of £2.3m per person annually

OpenAI UK reported 2025 turnover of £293m, more than triple the previous year, and a £20.3m profit; 110 staff shared £87m in salaries and £168m in company shares, averaging £2.32m per person. The article also cites AI safety incidents and a departing safety leader’s criticism of the industry, while describing intense competition for AI talent and rising San Francisco rents.

Analysis

The investable signal is not the headline compensation average; it is that AI talent is becoming a scarce, globally auctioned input. That can pressure incremental R&D costs and raise retention risk across London’s AI cluster, including Google DeepMind. For Alphabet (GOOG), this is a competitive-cost headwind, but the article gives no evidence of departures, pay changes, or a material effect on consolidated earnings. Meta (META) faces similar labor-market exposure; any reported recruiting packages should be treated as a cost-risk indicator, not proof of an immediate earnings hit.

OpenAI UK Ltd’s local revenue and profit do not establish group-level economics or the profitability of its research operation: the entity also houses sales and government-affairs functions, and the article does not explain cost allocation or equity-compensation accounting. The average package is also likely skewed by a small number of senior researchers, so extrapolating it to the wider workforce would overstate broad-based labor inflation.

Over 1–3 months, watch for disclosed hiring/retention costs, London researcher moves, or revised AI investment guidance—not this article alone—as catalysts for GOOG/META relative performance. Over 6–18 months, a sustained talent auction could modestly weigh on returns from AI investment; conversely, stronger monetization or productivity gains would absorb the cost. Safety incidents create a two-sided risk: tighter scrutiny could raise compliance and deployment costs, while credible safeguards could differentiate established platforms. Contrarian point: high compensation is not itself evidence of waste; in a winner-take-most market, scarce researchers may be economically valuable. No clear standalone trade signal here.

Key Decisions for Investors

• No immediate position based solely on this report. Treat the compensation figures as a labor-cost and talent-retention watch item, not a quantified earnings shock to GOOG or META.
• For the next 1–3 months, monitor Alphabet and Meta earnings commentary for AI-related personnel expense, recruiting/retention disclosures, and evidence that AI products are monetizing fast enough to offset rising development costs.
• A relative long GOOG / short META is only a conditional watch, not a recommendation: consider it only if evidence emerges that Meta is paying materially more to retain or recruit talent without corresponding monetization progress. Falsify the thesis if META demonstrates improving AI-driven monetization or GOOG signals worsening DeepMind retention or investment returns.
• Track safety-related regulatory actions and product restrictions over 6–18 months. A material constraint on deployment would weaken the AI investment case across both companies; evidence of incidents being contained without tighter operating limits would reduce that risk.