8 Comments
User's avatar
nVariant Capital Fund's avatar

I approached it from a slightly different angle, with similar conclusions.

https://nvariant.substack.com/p/is-space-based-ai-hosting-a-scam

Hazelnuts Research's avatar

All roads lead to the same conclusion- great write up and really enjoyed reading it!

Neural Foundry's avatar

Outstanding breakdown on the thermal bottleneck for orbital datacenters. The Stefan-Boltzmann constraint is something most hype pieces conveniently skip over, but radiative cooling at scale isn't just expensive, its geometry alone makes these systems impractical. I've seen similar overconfidence in edge deployments where power density assumtions fall apart once you account for real-world cooling limits. The camera-only robotaxi approach feels like anothr case of optimizing for demo conditions rather than operational robustness.

Hazelnuts Research's avatar

Exactly! A vision has to be built on a base but what exactly is the base here? Thank you so much for taking the time and reading it!

Semih Öztürk's avatar

Great article. The focus on physical infrastructure is crucial. While software is often quickly replicable (the 'race to the bottom'), hardware development for humanoids or space infrastructure presents a massive barrier to entry. This is where the wheat will be separated from the chaff, as these products are far more complex both physically and technologically. It will be fascinating to see who masters the scaling process. Which hardware verticals do you consider to be the most critical over the next 2-3 years?

Hazelnuts Research's avatar

Definitely AV’s since waymo is expanding rapidly while robotaxis stall due to their faulty nature. Thanks for reading!

Sensus Capital Research's avatar

Great report! I like the new flow and style.

Hazelnuts Research's avatar

Thank you so much!!