🌊 The rarest asset in venture

AI bubbles, defence booms, trillion-dollar charts, and courage > consensus.

I’m Ivan. This is a weekly brief on how top 1% startups use ai, raise capital, and grow. The highest signal-to-word startup newsletter you’ll read.

---

This week’s sponsor is AI CRM Attio:

Attio is the AI-native CRM for the next era of companies: Connect your email and calendar, and Attio instantly builds a CRM that matches your business model — with all your companies, contacts, and interactions enriched with actionable insights. With Attio, AI isn’t just a feature - it’s the foundation. You can do things like:

• Instantly prospect and route leads with research agents

• Get real-time insights from AI during customer conversations

• Build powerful AI automations for your most complex workflows

---

Hello there - I’m back from SF and (very) jet-lagged. But it was totally worth it.

Here’s what went down in startup-land this week:

🤖 AI

The FT shared intel on OpenAI’s trillion-dollar web. There’s a big debate on bubble vs no bubble. I wrote a piece arguing that the answer is both. If you zoom in we have a bubble in the making. Financial crises are endogenous to capitalism, here’s Mr Dalio backing me up. On the other hand, if you zoom out and look at a 10 year horizon, we may be at the beginning of a tech supercycle.

New OpenAI performance numbers and their (interesting?) ownership structure popped up. My take is that a company this size with so much riding on it should just go public. Not just because this “lets stay private forever” playbook distorts early-stage markets, but also because it should be treated like any other major institution shaping society (more transparent, accountable, governance, etc.).

Found some great performance data on the most valuable private AI startups. Valuation multiples will make you dizzy. But so will their growth rates (Cognition, Cursor, Anthropic).

CB Insights published a list of the Top 20 AI agent startups that have crossed $10M in ARR. Six of the top 20 build for developers and half of them are <3 years old.

The State of AI report dropped, the tl-dr: OpenAI still leads the frontier but the gap is closing fast. 2025 was the year AI learned to think: plan, reflect, and self-correct. Adoption went mainstream (44% of U.S. companies now pay for AI), and compute is the new bottleneck (we touched on this last week).

Meta just dropped another big AI paper (Agent Learning via Early Experience) and it’s a glimpse into how agents will actually learn in the wild. Instead of relying on curated human data or slow reinforcement loops, the model learns by doing, running its own “what if I do this instead?” simulations to improve reasoning. Meanwhile, Zuck keeps vaccuuming ai talent (Andrew, cofounder of Thinking Machines Lab).

I shared a short history of LLMs, with a phenomenal dataset and visual by Harrisson Schell, 8 years of LLMs compressed into one curve of momentum.

🌊 VC

I loved this article from Nicholas Chirls arguing that courage, not intellect, is now the rarest edge in venture. Today everyone reads the same books, listens to the same podcasts, and runs the same AI tools. One of the only true differentiators left (apart from subscribing to Startup Riders Pro 😎) is having the guts to bet when it feels hard (non-consensus). Basically to risk your reputation, not just your capital.