A $986,000 Google salesperson quit to build an AI startup. The equity gap is why the next one will too.

At 41, Yousuf Imran was earning about $986,000 a year selling advertising for Google. He left in April.

Imran, a Bay Area account executive, was pulling in roughly $170,000 in base pay plus commissions when he walked out to launch a startup building AI sales tools, a business he had been sketching in his own time using his employer's infrastructure. The deciding factor, he told Business Insider, was not base salary. "Google pays very well, but the equity packages at OpenAI and Anthropic are in a different universe."

The compensation gap Imran described has become a self-reinforcing feedback loop at Google. Each high-profile departure resets the comp benchmark inside the company, makes the next recruiter pitch easier, and lowers the social cost of leaving. The labs, not Google's own recruiters, are running the loop.

Twelve current and former Google employees spoke to Business Insider; six had already left. Their reasons cluster around three forces, all of which feed the loop. Google refresh grants have not kept pace with the appreciation of AI-lab equity held by early employees. Post-2023 return-to-office mandates have sapped the residual loyalty that used to compensate for slower comp growth. And the engineers and salespeople most exposed to AI in their day jobs are the ones most aware they are working on yesterday's product.

OpenAI and Anthropic together hired roughly 100 Salesforce employees this year, per Benzinga, and Fortune reported in June 2025 that Anthropic was already winning a one-sided talent war against OpenAI and DeepMind for senior AI engineers. The labs are repricing the senior labor pool for technical work across incumbents, not poaching from any single one.

OpenAI plans to nearly double its workforce to about 8,000 by the end of 2026, Reuters reported, citing the Financial Times, an absorption rate of more than 3,000 net hires over roughly twenty months. Anthropic's 2026 leadership bench, per techfundingnews, is dominated by executives recruited from OpenAI, Google, and Stripe. There are seats for the people Google is losing.

AI-lab equity packages vest faster and dilute existing shareholders less than Google grants do. On the secondary market, that structure lets them outpace a Google refresh grant within a single vesting cycle. Google's compensation model assumes predictable refresh grants and a base salary high enough to retain middle-of-the-bell-curve employees. The model breaks when the relevant benchmark for a senior IC is the price of a private tender offer at Anthropic, not a Google RSU refresh table. Closing the gap would force Google to either dilute faster than its buyback program can absorb, or accept a step-change in compensation cost the rest of the workforce would then demand.

The closest industry benchmark on this dynamic is the SignalFire State of Tech Talent Report 2025. SignalFire is a venture firm with a recruiting arm and a commercial interest in the conclusion; the data should be read as an industry view rather than a neutral survey. Its directional finding is consistent with the Business Insider reporting: retention has weakened most sharply at the largest tech employers among employees whose work is closest to AI, and most of that movement lands at the labs rather than at a big-tech competitor.

The senior engineers and senior sellers leaving Google, like Imran, are people with direct experience of the integration friction between AI models and the products they slot into. On the BI interviewees' account, the labs are buying that practical knowledge when they poach from incumbents. Each movement concentrates the operational know-how of how AI ships inside the labs, where the people who built the models are already concentrated. The result is a tighter feedback loop between research and product than any incumbent, Google included, can match with a remote hire.

OpenAI's stated end-of-2026 headcount target is 8,000. If the company overshoots, the absorption rate has accelerated beyond what the Financial Times reported in March. If it undershoots, the equity gap is widening faster than the labs can hire against it. Either reading is a dated signal about whether the feedback loop Google is running against is strengthening or peaking.