Darrick Horton, TensorWave | theCUBE + NYSE Wired: Media Week - Cyber & AI Innovators Summit
summary
A GPU cloud provider builds AI-focused data centers using AMD alternative to NVIDIA hardware and an emphasis on stack optimization and bespoke customer integrations. Funding shows a $44M seed via SAFEs with a priced Series A planned in mid‑nine figures and heavy use of collateralized debt. Infrastructure totals 1+ GW under contract, current sites at ~50 MW, rack densities of 120–200 kW per rack, widespread liquid cooling, and a 2025 roadmap to scale toward hundreds of thousands of GPUs.
AI-written summary · about
chapters (13)
- 0:00Introduction and welcome
- 0:24About TensorWave
- 1:29Competing with hyperscalers
- 3:14AI vs general workloads
- 4:41GPUs for inference debate
- 5:46Business model and customers
- 7:14Why AMD partnership
- 9:40Funding and scaling data centers
- 11:08Market for AI workloads
- 14:45Energy and data center power
- 16:31Liquid cooling in AI
- 20:33NYSE closing bell moment
- 20:54Future plans and scaling
