Open vs. Closed AI: How Founders Decide What to Build On at TechCrunch Disrupt 2026

Meta description: TechCrunch Disrupt 2026 sessions trace how AI startups choose, customize, and mix models across cost, ownership, and hardware from October 13-15.

Tags: TechCrunch Disrupt 2026, AI model choice, open models, proprietary APIs, AI hardware


TechCrunch Disrupt 2026 is shaping a timeline for AI builders around a practical question: how companies decide which models to use, own, or mix as capabilities shift. The event runs October 13-15 at Moscone West in San Francisco and gathers more than 10,000 founders, investors, operators, and tech leaders across 200+ sessions on six industry stages, roundtables, and breakouts. Its four conversations approach those decisions from different points in the AI stack: multi-model applications, customized models, infrastructure, and the chips underneath them. For startups, the agenda reflects a broader change: model selection is no longer a single architecture decision but a sequence of choices about cost, control, and hardware.

Background: Model Choice Becomes a Continuous Decision

Earlier AI stacks often forced teams to commit to one path: an open model, a proprietary API, or a custom build. By 2026, that framing is less stable. Open models are improving, frontier APIs, or hosted large-model services, keep advancing, and customized models can serve specific workloads. Some products now call on multiple models rather than betting on one. That gives founders more ways to build, but it also creates decisions about where to spend, what to own, and how much flexibility to preserve as the technology changes.

Development: Disrupt 2026 Sessions Map the AI Stack

TechCrunch Disrupt 2026 organizes those decisions across four stages of the stack. On the Builder's Stage, Mo Jomaa, partner at CapitalG, Vipul Ved Prakash, co-founder and CEO of Together AI, and Zuzanna Stamirowska, CEO and co-founder of Pathway, host "The Real Tokenmaxxing: How the Best AI Companies Navigate a Multi-Model World." The session looks at why companies are using multiple models, how they balance cost, performance, and flexibility, and when open models can outperform proprietary alternatives.

On the Real World AI Stage, Manos Koukoumidis, CEO and co-founder of Oumi, presents "Which AI Should Your Company Actually Deploy: Rent, Customize, or Build." He will use audience polls, startup scenarios, and a practical framework to compare frontier APIs, customized open weights, and owning AI outright. Attendees will come away with three decision principles they can bring to their next architecture conversation. Owning more of that stack may bring control and differentiation, while also demanding more time, talent, and resources than using existing models.

On the Builders Stage, Nader Khalil, Director of Developer Tech at Nvidia, and Sydney Sykes, Global Head of VC Partnerships at Nvidia, discuss "Building AI Startups Worth Betting On." Their conversation examines what founders are choosing today, the trade-offs between frontier APIs and open-weight models, or models whose parameters can be deployed by companies, and how those choices can affect product strategy and long-term differentiation. Model choice can shape costs, infrastructure requirements, and how much control a startup has over its product. It can also determine where, and how easily, a startup can create differentiation competitors cannot replicate.

On the Disrupt Stage, Anna Goldie, founder and CEO of Ricursive Intelligence, and Azalia Mirhoseini, founder and CTO, address "When AI Starts Designing Its Own Hardware." They explore how AI is optimizing chips and hardware, why model architecture and hardware are becoming more closely connected, and what an increasingly open AI ecosystem could mean for the infrastructure underneath it. For founders, the connection between AI and hardware could affect how quickly new capabilities reach the market. Faster chip development could also change the infrastructure available to startups building the next generation of AI products. Image credits in related materials include Together AI, Slava Blazer Photography / Flickr, and Ricursive Intelligence.

Latest Progress: Passes, Discounts, and Event Scale

Disrupt 2026 is also positioned around access. Passes offer savings of up to $100, with a second pass at 50% off. For a limited time, an Expo+ Pass costs $75 for people affected by a layoff. The event expects more than 10,000 attendees, 250+ speakers, and 300+ exhibiting startups. Matchmaking, dealmaking, and ad hoc networking are intended to connect founders, investors, potential partners, and other builders facing many of the same technology and business decisions.

Current Stage: Flexibility Over Single Commitment

The current stage of AI product building is less about picking a permanent model and more about managing transitions. A startup may use a frontier API today, customize an open model tomorrow, or eventually move workloads among several models. That flexibility can shape operating costs, product decisions, and how quickly a company can adopt better models as they emerge. Preserving flexibility could be as important as choosing the right model now.

What to watch next is how these sessions turn those choices into operating practice: multi-model routing, ownership boundaries, builder and investor perspectives from Nvidia, and hardware designed with AI.