Issue #138 • September 8, 2026

Hi Seedradar Squad! Capital is flooding straight into raw physical infrastructure. Crusoe locked down $3B for power-dense data centers, Nscale is rounding up $3.5B with Nvidia in its corner, and XDOF hit a $1.2B valuation in just 90 days strictly by training robots how to move. The pipes are getting massive, and the race to control physical hardware has completely taken over the landscape.

At the same time, the focus is shifting away from generic software wrappers toward leaner, sharper execution. That's why Deep Cogito is our star pick this week with their open-weight hybrid reasoning, right alongside huge bets on Nexthop's open networking gear and Gimlet's multi-silicon inference cloud. The playbook is set: stop chasing buzzwords and back the actual architecture running underneath.

Deep Cogito, an AI startup founded by former Google engineers, just pulled in a $43 million Series A to build open-weight hybrid reasoning models. Instead of forcing a one-size-fits-all approach, their systems seamlessly toggle between firing off fast direct answers for simple queries and engaging in deep, step-by-step reasoning for complex problems.

They aren't just training another generic chatbot from scratch. By applying novel post-training techniques and reinforcement learning to existing open-source foundations, they are pushing toward iterative self-improvement. This hybrid architecture gives developers the flexibility of two distinct modes within a single deployment, putting the actual horsepower back into the hands of engineering teams.

-Learn More About Deep Cogito
-Learn More About Deep Cogito’s funding round

At Seedradar Ventures, we back relentless founders at the earliest stage, investing in teams solving big problems with solutions that deliver a 10–20x greater impact.

Our syndicate offers investors access to a curated pipeline of early-stage startups, sourced through our network.

👉 Ready to learn more? Explore the Seedradar Ventures Syndicate on AngelList