Four sessions at TechCrunch Disrupt 2026 take on model choice, examining distinct layers of the AI stack. Together, they show a change in how AI firms approach their building blocks. Instead of locking into one model for good, founders now see model selection as something that can be changed over time.
Moscone West hosts the event from October 13 through 15, where over 10,000 founders, investors, operators, and tech leaders will gather in San Francisco. The program fills six stages, roundtables, and breakouts, featuring more than 250 speakers and 300 exhibiting startups.
The Multi-Model Approach
The Builder’s Stage hosts “The Real Tokenmaxxing: How the Best AI Companies Navigate a Multi-Model World.” Mo Jomaa, partner at CapitalG, joins Vipul Ved Prakash, co-founder and CEO of Together AI, and Zuzanna Stamirowska, CEO and co-founder of Pathway, to discuss why companies are using multiple models.
Founders weigh the trade-offs between cost, performance, and flexibility when picking a model, and wonder whether open models can beat proprietary ones. The session poses these questions, and the answers carry weight far beyond model performance alone. Choices about operating costs, product decisions, and the pace of adoption all rest on the model selections made.
Owning the Stack
Manos Koukoumidis, CEO and co-founder of Oumi, takes the Real World AI Stage for “Which AI Should Your Company Actually Deploy: Rent, Customize, or Build.” He will cover whether startups should build their own models, when customization becomes a competitive advantage, and how to choose between open and proprietary AI.
Koukoumidis is bringing audience polls, startup scenarios, and a practical framework to help people figure out whether to go with frontier APIs, customized open weights, or owning AI outright. The goal is for attendees to walk away with three decision principles they can put into practice during their next conversation about architecture.
API Trade-offs
Sydney Sykes, who leads VC partnerships globally at Nvidia, and Nader Khalil, Nvidia’s director of developer tech, step onto the Builders Stage for “Building AI Startups Worth Betting On.” The pair will look into what founders are selecting right now, weigh the pros and cons of frontier APIs against open-weight models, and consider how these decisions shape product strategy and long-term differentiation.
The selection of a model influences expenses, the necessary infrastructure, and oversight. It can also decide where a startup can establish a distinctive advantage that competitors cannot copy.
AI Designing Its Own Hardware
Ricursive Intelligence’s founders Anna Goldie, CEO, and Azalia Mirhoseini, CTO, will appear on the Disrupt Stage in “When AI Starts Designing Its Own Hardware.” Their talk will cover AI’s role in chip optimization and hardware design, the growing bond between model architecture and hardware, and the implications of an increasingly open AI ecosystem for the underlying infrastructure.
The infrastructure available to startups building the next generation of AI products could be changed by faster chip development.
The Full Lineup
Here is the full lineup of sessions:
| Session | Speakers | Stage |
|---|---|---|
| Tokenmaxxing | Jomaa, Ved Prakash, Stamirowska | Builder’s Stage |
| Which AI Should Deploy | Koukoumidis | Real World AI Stage |
| Building AI Startups Worth Betting On | Khalil, Sykes | Builders Stage |
| When AI Designs Its Own Hardware | Goldie, Mirhoseini | Disrupt Stage |
Key Numbers
- Conference: TechCrunch Disrupt 2026
- Dates: October 13–15
- Location: Moscone West, San Francisco
- Expected attendees: more than 10,000
- Speakers: 250+
- Exhibiting startups: 300+
- Save up to $100 on a pass
- Second pass at 50% off
- Expo+ Pass at $75 for those affected by a layoff
What Founders Need to Know
Today, a startup can lean on a frontier API, tailor an open model later, or shift workloads across multiple models over time. Keeping options open might prove just as crucial as picking the correct model at the outset. Attendees get to learn from founders who have already navigated these decisions, via polling, scenarios, and hands-on frameworks.
Six stages, roundtables, and breakouts make up a larger set of 200+ sessions, and within them, matchmaking, dealmaking, and spontaneous networking offer attendees ways to meet founders, investors, possible partners, and other builders who face similar technology and business choices.
Source material: “Open or closed AI? How founders are choosing what to build on at TechCrunch Disrupt 2026,” TechCrunch.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.