Nitski's AI Hot Take: Services First, Products Later
Mercor's CPO Oswald Nitski unveils his AI deployment endgame: services now, products later. Plus, how to win SF's brutal talent war by prioritizing agency.
Mercor Head of Product on Revenue Concentration from Frontier Labs
With Harry Stebbings, Osvald Nitski · Sunday, July 26, 2026
In this episode, Oswald Nitski, CPO at Mercor, discusses the complex relationship between open-source models and data providers, arguing that open AI raises the floor but true value lies in specialized frontier data for latent demand. He delves into enterprise AI ROI, the shift towards hyper-specialized models, and the transformative impact of AI on product management, emphasizing judgment and simplification. Nitski also details Mercor's strategy for scaling human data annotation, navigating revenue concentration with frontier labs, and offers insights into the future of AI deployment, including the burgeoning field of robotics and the brutal San Francisco talent market.
Key takeaways
- AI deployment will be services-heavy in the short term: Oswald Nitski, CPO at Mercor, predicts a significant reliance on AI services because companies currently lack the in-house knowledge to deploy models themselves. Read more →
- Mercor CPO Oswald Nitski insists that the perceived ROI problem for enterprise AI isn't a problem at all. Instead, it's a necessary period of exploration and experimentation where companies have more patience for returns. Read more →
- Open-source AI models raise the general capability floor but don't cannibalize the value of specialized, frontier data for companies like Mercor. Read more →
3 articles from this episode
AI ROI Isn't a Problem, It's Exploration: Mercor CPO
Mercor CPO Oswald Nitski argues enterprise AI isn't failing ROI, it's exploring. Get ready for hyper-specialized models and rising AI spend beyond 3% of developer salaries.
Open Source Doesn't Kill Data Value, It Reveals Latent Demand
Oswald Nitski of Mercor explains how open-source AI raises the floor but specialized data uncovers new, latent enterprise demand, despite client data fears.
More 20VC with Harry Stebbings episodes
- Modal CEO: Has Open-Source LLM Inference Peaked? | What Happens to the Cost of ComputeNov 8 · 4 articles
- Instinct Worth 30BN? | Microsoft Cuts 27,000 Jobs While Meta Taps Equity Markets for $95B CapExNov 1 · 5 articles
- Satya Nadella: Whether to Trust Frontier Models | Why Companies Need to Own Their Own IntelligenceNov 1 · 5 articles
- David Tisch: Why No Manager Wants to Stay Small & Why Competing with Mega Platforms is a Losing GameNov 1 · 6 articles