Q&A with Nvidia CEO Jensen Huang, who says the company is now working with 4,500 AI startups, compared to 2,000 in April, 2017
Dean Takahashi / VentureBeat :
Context & Ripple Effects
This 2019 VentureBeat Q&A lands mid-arc in Huang's long run of interviews: Fortune's profile two years earlier had him holding roughly 70% of the GPU market as AI demand exploded, and this conversation quantifies what that demand looks like from the inside — the AI startup count Nvidia works with has doubled since April 2017 to 4,500.
Read against later coverage, the number marks the moment Nvidia's moat shifted from chips to ecosystem: by 2022 Huang was framing the company around a full-stack hardware-plus-software approach, and by 2024 Fast Company was asking why that ubiquity will not be easy to replicate.
First-order effects
- Nvidia's direct beneficiary count doubles on paper: 4,500 AI startups are now building against its tooling and silicon, each one adding switching costs to the installed base Fortune measured at 70% GPU share.
Second-order effects
- Rivals competing for the same developers can no longer win on hardware specs alone — the Stratechery interview's integration-of-hardware-and-software thesis becomes the forced playbook for anyone attacking Nvidia's share.
Third-order effects
- If the startup-funnel pattern holds, the industry consolidates around whoever owns the developer relationship: the Wired interview's 'AI factory' data-center concept is the endpoint of that logic, where compute, software, and customers are bundled into one platform rather than sold as parts.
The trend: AI compute competition is shifting from chip-level market share to ecosystem capture, with Nvidia converting a doubling startup base into durable full-stack lock-in.