Should enterprises fear becoming dependent on frontier model providers for their core work?

Harry Stebbings · Nov 22, 2023

The largest enterprises say they will not send customer and transaction data to a model outside their own bounds

Scope: based on his conversations with large enterprises

29:25 20VC: Why OpenAI Will Become an Infrastructure Play, Why Apple Will Win in an AI World, Why Google is the Most Vulnerable Incumbent, Will LLMs Be Commoditised, Which Startups Are Thin vs Thick Wrappers on Top of LLMs with Jeff Seibert, Founder @ Digits

Arvind Jain · Jul 11, 2026

Enterprise leaders are genuinely terrified of frontier model providers, fearing their core IP, data and ways of working become dependent on those providers

Like every software company, enterprises worry the models will eat everything and create excessive technology dependence

6:47 20VC: Why OpenAI and Anthropic Won't Win the App Layer | Why Teams Will Get Bigger Not Smaller in a World of AI | Why AI Removes Incumbents Advantage of Bundling | China vs America: Who Wins the AI War with Arvind Jain, Co-Founder @ Glean

Alexandr Wang · Jun 12, 2024

Serious large enterprises will move toward on-prem open-source models they customize on their own data, creating a huge unmet opportunity for Llama, Mistral and similar models

Enterprise leaders now realize their proprietary data may be their only competitive differentiator in an AI world, and sharing it with a model developer risks mortgaging their entire future; they need hard guarantees their data won't improve competitors

22:45 20VC: Scale's Alex Wang on Why Data Not Compute is the Bottleneck to Foundation Model Performance, Why AI is the Greatest Military Asset Ever, Is China Really Two Years Behind the US in AI and Why the CCPs Industrial Approach is Better than Anyone Else's

Anastasios Angelopoulos · Aug 3, 2026 · hedged

Forward-deployed engineering is not unique, but combining FDE with an open American model may be a more sustainable business for Western enterprises than depending on frontier providers.

Enterprises may want to cut out third-party services for cost and sovereignty reasons, own and continually fine-tune the model themselves, and avoid supply chain risk.

Scope: 'may be' — not asserted as certain

12:38 20VC: 70% of Neolabs Will Die | There Will be a $100BN US Open-Source Model | Data is a Trillion $ Market | Governments Cannot Regulate Models: It is Too Late | The Cyber Attacks to Come Will be Insane with Anastasios Angelopoulos @ Arena

Jeff Seibert · Nov 22, 2023

Founders shouldn't fear depending on external LLM providers — LLMs are a tool like databases, just evolving faster; what matters is speed of adaptation and staying focused on the customer's problem

It's technology, not life or death for the company; the real requirement is being able to adapt to the new reality faster than competitors, which weekly sprints enable

20:48 Llms are just another tool like databases dependency fear is overblown speed of adaptation matters more

Anastasios Angelopoulos · Aug 3, 2026 · hedged

Enterprises both fear the frontier labs as competitors and embrace them as vendors — you see both dynamics in the market simultaneously

52:18 Enterprises simultaneously fear labs as rivals and buy from them

Arvind Jain · Jul 11, 2026

Enterprises are right to fear frontier model providers, because if agents run by those providers do the majority of work, the enterprise transfers real operational control and loses ownership of its compounding institutional learning

Most process knowledge is never documented and accumulates through repetition; in future that learning accumulates inside the agent, so whoever runs the agent owns the learning

7:32 Agents accumulate the institutional learning so whoever runs them owns it