Open Models Change The Economics of AI

Name: Open Models Change The Economics of AI
Uploaded: 2026-09-04T14:00:10.000Z
Duration: 57 min 16 s
Channel: Y Combinator

Y CombinatorSeptember 4, 202657m

Topics63

Open Models Address Cost and Customization in Enterprise AI0:00 State of AI with Jeffrey Morgan of Ollama1:03 Token Flow Data and Usage Trends1:30 Chinese Models Dominate Cloud Usage2:00 Cost as Entry Point to Customization2:00 AT&T's 40% Shift to Open Models2:30 Coding Agents and OpenClaw Drive Token Usage3:01 Exponential Token Usage Growth3:30 Context Window Expansion Enables Growth4:30 Out-of-the-Box Open Models Take Off5:02 Model Release Cadence Accelerating5:30 AI Safety Creates Opportunity for Open Models6:30 Open Models Excel at Security Testing7:30 Ollama's Role in Model Launches8:30 Deepseek's First Multimodal Model9:00 Ollama Creates Legible Standards10:00 Packaging Models with Hardware and Providers10:30 Ollama as Operating System for AI11:30 Building Technical Expertise12:33 Reproducing the Five-Layer Cake for Open Models13:00 Hidden Layers in the Stack13:30 Unbundling Opportunities14:00 Bundling vs Unbundling Dynamics15:00 Coding Agents Reducing Lock-in16:00 Memory as Stateful Problem16:30 Future State: 80-90% Open Models17:30 Cloud Computing Parallel20:00 Local Models Mix with Cloud21:00 Hybrid Execution Model for AI Workloads22:08 Geographic Distribution of Model Usage23:01 Nvidia's Open Source Strategy24:00 Desktop AI Hardware Pricing24:31 DGX Spark Capabilities25:32 MLX Technology Stack26:30 Local vs Cloud Coding Experience27:00 GPU Market Dynamics27:30 Developer Access Solutions28:30 Startup Budget Optimization Strategy29:00 Efficiency and Adoption Patterns30:00 Model Orchestration Benefits31:00 Model Architecture Philosophy31:31 Frontier Competition Landscape33:00 Geopolitical Considerations34:00 Security and Model Integrity35:00 Ollama Origins and Founding Story36:30 Naming and Branding38:30 Series A Pitch and Investment39:00 Pre-Pivot Struggles40:01 Critical Pivot Decision42:01 Initial User Reception43:30 Early Days and Initial Adoption44:22 Rapid Growth and Product-Market Fit45:00 Speedrun from Homebrew to Enterprise Adoption46:01 Monetization Timeline and Business Model Development47:00 Strategic Patience and Market Timing48:00 Risks of Waiting and Customer Connection49:01 Decision to Join YC and Second-Time Founder Perspective50:01 Learning from Previous Companies and Team Experience51:31 Breaking Old Infrastructure Analogies53:32 Non-Determinism as a Feature54:30 Building Services with Unknown Code55:01 Curation and Integration Value56:00 Closing57:00

In a Nutshell

Open models are now economically viable for enterprise AI because they solve the primary blocker of cost, with Chinese models dominating cloud usage and a 40% shift already seen at companies like AT&T. Token consumption is exploding 10-150x due to coding agents and coworker-style automation enabled by expanded context windows, driving demand for efficient orchestration layers above the models themselves. The long-term equilibrium is 80-90% open model usage with hybrid local/cloud execution, where frontier models handle only the hardest tasks while open models power the majority of work through specialized routing and coordination systems.

AI-Generated Notes

These notes were generated by AI and may contain inaccuracies.

Cost is by far the largest pain point that open models can jump in and solve. Every business has a vision of getting better control over AI and customizing it for their business, and that's really their north star. Cost is something they can solve in the short term, but it then enables them to customize these models for their unique use case.

Early 2024, there was lots of interest in fine-tuning your own custom models. Then it sort of went away and all of that it will just be wasted effort. It'll get stomped by the next model release. It seems like it's coming back now.

Ollama is used by 9 million developers, has 178,000 GitHub stars, and is used by 85% of the Fortune 500. Jeffrey Morgan, co-founder and CEO of Ollama, has a front seat to what models score highest on benchmarks and what developers actually download and keep using.

The biggest thing being seen is a shift to open models, especially in enterprise. This is from a mix of US and Chinese origin models, and it's predominantly driven by coding agents and also AI assistants more coworker cases like OpenClaw and Hermes.

Ollama sits in the token flow of so many tokens and has really good data on what models people are actually using and how it's changing. Ollama started as a way to run open models on MacBook or other hardware including Nvidia, AMD, Intel. Earlier this year, Ollama launched Ollama's cloud.

What is being seen on Ollama's cloud is that it's predominantly Chinese models right now of Chinese origin. But they're being accessed by businesses all over the world, especially US and Germany, which is actually a big source of where open model tokens are being accessed.

Cost is by far the largest pain point that open models can jump in and solve, but every business has a vision of getting better control over AI and customizing it for their business, and that's really their north star. Cost is something they can solve in the short term, but it then enables them to go and customize these models for their unique use case.

Sign in to read the full notes

Get access to AI-generated notes, topic timestamps, and more.

Open Models Change The Economics of AI

Open Models Address Cost and Customization in Enterprise AI00:00

State of AI with Jeffrey Morgan of Ollama01:03

Token Flow Data and Usage Trends01:30

Sign in to read the full notes