Openai bucks lay-off trend with plan to double headcount by 2026

While rivals hand out redundancy notices and blame generative ai for the savings, OpenAI is writing the opposite script. The company that ignited the current boom now aims to swell from 4,500 to roughly 8,000 employees before the 2026 holiday parties begin, according to people familiar with the hiring roadmap and first reported by the Financial Times.

The arithmetic is brutal everywhere else. Meta just trimmed another 1,500 roles; Amazon has erased 30,000 desks since October. Their shared boardroom slide: ai automation. Fewer coders, fewer ad-sales foot soldiers, fewer humans. OpenAI’s retort? Bring in more.

Recruitment sprint targets product, research and enterprise sales

Headcount will land first in product engineering, the research bench that keeps GPT ahead of open-source clones, and a beefed-up enterprise sales unit tasked with showing Fortune 500 clients how to squeeze margin out of models they already rent by the token. The message to CIOs: you won’t need a PhD in prompt craft, we’ll supply the experts.

To house the influx, OpenAI has signed a second San Francisco lease, stacking yet another glass box into the city’s half-empty skyline. Hybrid work is allowed on paper; the unspoken expectation is five-day presence. The return-to-office memo is arriving fashionably late, but it is arriving.

The hiring wave is not charity. Anthropic’s Claude 3 has been matching—and occasionally surpassing—GPT-4 on several benchmarks, and Google’s Gemini Ultra is circling. Talent is the most portable moat in this race. Every researcher poached is one the competition cannot have.

San francisco real estate tells its own story

Commercial brokers in SoMa whisper that OpenAI’s new footprint is the largest single lease signed this year, a lifeline to landlords still bleeding from the remote-work exodus. The symbolism is delicious: the firm that automates white-collar tasks now props up the very district built for white collars.

Not everyone inside the company is cheering. Some engineers grumble that doubling in two years reeks of the same blitz-scaling that bloated Facebook and Uber before their own reckonings. OpenAI counters with cash: revenue is reportedly approaching $2 billion annualised, most of it from ChatGPT Plus and enterprise tiers. When your cash register grows faster than your payroll, you can afford to hire first and ask questions later.

Still, the clock is ticking. Training costs for frontier models double roughly every six months; regulatory scrutiny is sharpening in Brussels and Washington; and GPU supply remains Nvidia’s game. Adding bodies does not guarantee breakthroughs, but falling behind on talent almost guarantees stagnation.

So the resumes keep pouring in, the recruiters keep pinging, and San Francisco’s cafés once again echo with talk of transformer architectures and stock-option strike prices. For an industry obsessed with automating labour, the hottest start-up has concluded that, for now, humans remain a growth industry. The lay-off narrative will have to wait.