Rezo🛡₿RRR

A day in the life of a VC in 2026: 9am: Board meeting. My main value-add is aggressively pushing for Anthropic/OpenAI usage in non-engineering functions. Briefly ponder how I became an SDR for foundation model companies. 1pm: Lunch with another VC. We discuss how startups can find "blue ocean" away Anthropic/OpenAI. We conclude we should probably just invest the rest of our funds directly into Anthropic/OpenAI. 3pm: Pitch meeting. Me: "Do you run on Anthropic or OpenAI?" Founder: "Both." Debating internally whether a company reselling Anthropic/OpenAI with a 10% gross margin is a good investment but hey, at least they're in the "token flow". 4:30pm: Deep due diligence. I ask Claude if it plans to build this exact startup natively in its next release. Same to ChatGPT. They both say yes. I pass on the deal. 6pm: Urgent call from a portco CTO: "We need an intro to upgrade our Anthropic tier!" I immediately agree to help them spend more of the venture dollars we just invested in them, on Anthropic. 8pm: Brainstorming next guests for the podcast. Thinking I should probably just try to get some folks from Anthropic and OpenAI.

LARRY ELLISON: AI IS RAPIDLY COMMODITIZING BECAUSE MOST MODELS ARE TRAINED ON THE SAME PUBLIC INTERNET DATA. THE REAL COMPETITIVE EDGE ISN’T THE MODEL ANYMORE — IT’S ACCESS TO EXCLUSIVE, PROPRIETARY DATASETS. THAT MAY BE THE ONLY MOAT LEFT.

The Cursor story is wild because the obvious takeaway is not the real one.... at a glance it looks like open-source model + 3 weeks of RL = suddenly you are competing with GPT-5.5, but that's not what happened. Composer v2.5 started from the same Kimi 2.5 base that powered Composer v2. Same model. Same starting point. The difference was not architecture but actual proprietary coding data, compute, and a feedback loop only Cursor could build. The base model was not the moat... the data was. This is exactly the thesis and problem I have been obsessed with at Super Protocol... not yesterday but years ago, even before there was a Cursor case proving it so clearly. Open-source makes intelligence cheaper, but the advantage moves to where the model cannot reach by itself so proprietary datasets inside companies, industries and workflows like: - Healthcare - Finance - Trading - Enterprise. Every serious industry has data that would make models much stronger, and none of them can simply expose it, upload it, or merge it with someone else’s data. That is where confidential computing stops being a privacy feature but the actual layer where proprietary data can collaborate without becoming public. Cursor is the proof. It just arrived after we had already started building the infrastructure for it. Data was never just an asset to keep in a vault... it was an advantage that could not be fully realized alone.

The Cursor story is wild because the obvious takeaway is not the real one.... at a glance it looks like open-source model + 3 weeks of RL = suddenly you are competing with GPT-5.5, but that's not what happened. Composer v2.5 started from the same Kimi 2.5 base that powered Composer v2. Same model. Same starting point. The difference was not architecture but actual proprietary coding data, compute, and a feedback loop only Cursor could build. The base model was not the moat... the data was. This is exactly the thesis and problem I have been obsessed with at Super Protocol... not yesterday but years ago, even before there was a Cursor case proving it so clearly. Open-source makes intelligence cheaper, but the advantage moves to where the model cannot reach by itself so proprietary datasets inside companies, industries and workflows like: - Healthcare - Finance - Trading - Enterprise. Every serious industry has data that would make models much stronger, and none of them can simply expose it, upload it, or merge it with someone else’s data. That is where confidential computing stops being a privacy feature but the actual layer where proprietary data can collaborate without becoming public. Cursor is the proof. It just arrived after we had already started building the infrastructure for it. Data was never just an asset to keep in a vault... it was an advantage that could not be fully realized alone.

The Cursor story is wild because the obvious takeaway is not the real one.... at a glance it looks like open-source model + 3 weeks of RL = suddenly you are competing with GPT-5.5, but that's not what happened. Composer v2.5 started from the same Kimi 2.5 base that powered Composer v2. Same model. Same starting point. The difference was not architecture but actual proprietary coding data, compute, and a feedback loop only Cursor could build. The base model was not the moat... the data was. This is exactly the thesis and problem I have been obsessed with at Super Protocol... not yesterday but years ago, even before there was a Cursor case proving it so clearly. Open-source makes intelligence cheaper, but the advantage moves to where the model cannot reach by itself so proprietary datasets inside companies, industries and workflows like: - Healthcare - Finance - Trading - Enterprise. Every serious industry has data that would make models much stronger, and none of them can simply expose it, upload it, or merge it with someone else’s data. That is where confidential computing stops being a privacy feature but the actual layer where proprietary data can collaborate without becoming public. Cursor is the proof. It just arrived after we had already started building the infrastructure for it. Data was never just an asset to keep in a vault... it was an advantage that could not be fully realized alone.