Jensen Huang’s GTC 2026 keynote wasn’t subtle. A $1 trillion order book. Thirty thousand attendees. A new supercomputing platform was unveiled via stage machinery – it was too heavy to lift by hand. The full scale of Nvidia’s AI empire was on display.
Beyond the showmanship, Huang spent considerable time explaining where the opportunities in the AI economy actually live and several of them are wide open for startups willing to move fast.
Tokens Are the New Product
Huang repeated a core concept at GTC: future data centers will no longer be places to store and process data, they will be “AI factories,” and their product is the token. That’s not just a metaphor for hyperscalers, is one we’re hearing from startups across our portfolio.
Startups that support domain-specific inference use cases, such as deploying models tuned to specific industries, workflows or data sets, are positioned to become the suppliers of the AI economy, not just its customers.
The Infrastructure Is Intentionally Open
Huang described the current moment as “a renaissance of enterprise IT,” comparing it directly to the arrival of HTML and Linux. Those platform shifts created entirely new categories of companies and the winners weren’t the giants, they were the fast movers who built early on open foundations.
Startups that move fast, spot the gaps and clear the infrastructure bottlenecks standing between enterprises and the AI economy won’t just participate in this renaissance, they’ll help define it.
Breaking The Cost Curve
For many companies adopting AI, the economics are brutal. Running inference at scale costs too much, which means the richest companies have a structural edge. While the Vera Rubin architecture claims to deliver ten times lower inference token costs and ten times more performance per watt than its predecessor, not every enterprise, neocloud or service provider has the capital to be an early adopter.
That’s where startups come in. Legacy infrastructure doesn’t retire overnight and the gap between what enterprises run today and where they’re headed is a durable revenue opportunity for startups that can help customers cut operating costs while the transition plays out.
The AI economy isn’t just being built by Nvidia, AWS and OpenAI, more than 240 Nvidia Inception startups showcased at GTC this year. The platform shift underway creates genuine opportunity for companies with limited resources and unlimited ambition. The ocean just got more interesting for the little fish who know how to move.
How you position your company within this shifting landscape, who you’re talking to, and what story you’re telling investors, customers and partners will be just as important as what you build. If you need help navigating, we’re here to help.