There are AI products to be built. Do you want to choose your own frontier model and get up and running? Do you want to build on an open model and have greater control? Do you tweak your version? Run it locally? Use multiple models? Will you change your strategy six months from now when economics and capabilities change again?
There may not be one correct answer. But for founders, making the wrong choice can impact just about everything that follows, including cost, infrastructure, profit margins, differentiation, speed, and control.
The decision will be at the center of the session “The Open vs. Closed AI Debate is Just Beginning” on the Builders Stage at TechCrunch Disrupt 2026, October 13-15 in San Francisco. This session, led by Nvidia’s Director of Developer Tech Nader Khalil and Global Head of VC Partnerships Sydney Sykes, will discuss the tradeoffs between open and proprietary AI and which approach can provide a lasting competitive advantage.
This is not a philosophical discussion about open source. This is a business decision that is currently being made within startups of all sizes. Get your pass and dive deep into AI discussions with 10,000+ technology leaders at Disrupt. Register now and save up to $200 before prices go up on September 25th at 11:59pm PT.
The AI gap is closing — decision-making is not getting easier.
The open model has advanced rapidly. Nvidia announced in July that its Nemotron open models and datasets were cited in 145 accepted papers at ICML 2026, alongside research using other Nvidia open model families across robotics, self-driving cars, and biomedical research.
At the same time, our own Frontier Lab continues to advance the model’s capabilities. As a result, the market is increasingly questioning whether open models are useful and instead asking what each approach makes commercially.
Even Nvidia rejects simple binary framing. At GTC earlier this year, CEO Jensen Huang argued that the future is not proprietary or open, but proprietary and open. It seems easy until you start a company based on this decision.
If two models can achieve similar results, will the lower cost win? What happens when you have more control over your data? Does owning more of the stack give you defensive power, or just the infrastructure you need to maintain? And if the best model changes every few months, how tightly should your product be tied to one of them? These are TechCrunch Disrupt 2026 This is the question that Khalil and Sykes unravel. Don’t miss it. Register for tickets now and save $200 by September 25th at 11:59pm PT.
Two perspectives shared by Builders Stage on the same AI stack
Nader Khalil approaches the debate from a builder and infrastructure perspective. Prior to becoming Director of Developer Technology at Nvidia, he led open source and local AI and co-founded Brev.dev, an AI infrastructure company that was acquired by Nvidia in July 2024.
Brev.dev was built around simplifying access to GPU infrastructure across a variety of environments. Nvidia’s developer documentation explains that it has tools that allow developers to deploy AI software across public clouds, private clouds, and on-premises infrastructure without being tied to a single compute source.
Meanwhile, Sidney Sykes is bringing the venture ecosystem to the table as NVIDIA’s global head of VC partnerships.
Taken together, these provide room to consider the same decision from different angles. What do developers need to build? What do companies need to become investable, scalable businesses?
Have a front row and center seat in one of the biggest debates in the world of AI today. Register for tickets before the up to $200 discount ends on September 25th at 11:59pm PT.
Your model is not your moat – until it is
Behind the open vs. closed debate lies another uncomfortable question. That’s where the competitive advantage actually lies.
If your competitors have access to the same proprietary API, you need to differentiate yourself elsewhere, such as through unique data, workflows, distribution, customer relationships, product experience, or specialized technology. However, choosing an open model does not automatically guarantee a moat.
It gives you flexibility and potentially greater control, but you also have to make decisions about deployment, optimization, and infrastructure. Also, economics can vary depending on workload and scale.
Nvidia is investing heavily in its open ecosystem. The company’s Nemotron 3 Super, launched in March, is a 120 billion parameter open model designed for agent workloads, and companies are already combining this with their own models rather than treating the two approaches as mutually exclusive. This hybrid reality may end up being the most interesting part of the discussion.
Register now to join the conversation on Builders Stage in October. Up to $200 off ends September 25th at 11:59pm PT.
What should you build on? Decided at TechCrunch Disrupt 2026
That’s why this session is useful beyond the AI engineer audience.
If you’re a founder, your choices can shape your profit margins, funding story, and product roadmap. If you’re an investor, understanding where value lies in the stack can help you distinguish between genuine defenses and thin product layers riding on someone else’s model. If you’re a line-of-business leader, this impacts procurement, security, data management, infrastructure, and the freedom to change providers later. It’s also an opportunity for developers and students to understand how the technical decisions being made today are directly related to the business models being built around them.
No one needs a more abstract debate about whether open or proprietary AI is philosophically better. What builders need is a clearer understanding of the trade-offs.
Join Nader Khalil and Sydney Sykes on the Builders stage at TechCrunch Disrupt 2026 and decide which side of your AI strategy you’re on. Register now and save up to $200 before prices go up on September 25th at 11:59pm PT.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
