Mercor, a cybersecurity company riding the AI wave, has grown quickly — so quickly that it's now confronting a stark reality: hiring top artificial intelligence talent costs tens of millions of dollars. Brendan Foody of venture firm 20VC described AI models as the core product driving Mercor's business, and said the price tag for the people who build them has become a central challenge.
The price of expertise
Foody, speaking on the 20VC podcast, said the market for elite AI researchers and engineers has turned into a bidding war. Companies like Mercor, he explained, are competing with tech giants and well-funded startups for a thin pool of specialists. The result is compensation packages that easily reach eight figures, particularly for those who have shipped large-scale models or published cutting-edge work.
That dynamic puts pressure on Mercor's bottom line. The firm's revenue growth is strong, but the cost of human capital eats into margins. Foody noted that businesses in this space have to weigh whether to hire full-time staff or rely on contract work — a decision that affects both speed and culture.
Client relationships as a safety net
When a crisis hits — a data breach, a model failure, a regulatory inquiry — strong client relationships become the difference between survival and collapse, Foody pointed out. Mercor has invested heavily in account management and transparency, making sure customers understand exactly what the AI is doing and why. That trust, he said, has kept retention rates high even as competitors undercut on price.
In cybersecurity, where mistakes can expose sensitive data or disrupt critical infrastructure, clients need to know their vendor can handle the worst. Mercor's approach has been to surface problems early and escalate fast, rather than hide them. Foody described that as a competitive moat that isn't easily copied.
AI models at the core
Foody was explicit: Mercor doesn't sell software or services in the traditional sense. It sells AI models. Those models are trained on proprietary data and tuned for specific security tasks — detecting intrusions, predicting vulnerabilities, automating responses. The product is the model, and the model is the product. That means Mercor's technical team isn't a support function; it's the engine.
Mercor's growth demonstrates AI's transformative impact on cybersecurity and the future job market, Foody argued. As more companies shift from rule-based systems to neural networks, the demand for model builders will only increase. But so will the cost. The question for Mercor — and for any firm betting on AI — is whether the revenue from those models can outrun the expense of the people who craft them.
Mercor hasn't announced its next funding round or hiring plans. Foody said the company is focused on scaling its client base before making another big talent push. Whether that strategy holds up under the pressure of a talent war remains to be seen.
