You'll own how Luma acquires compute - the sourcing strategy, the vendor relationships, and the multi-million-dollar contracts that put frontier-scale capacity in the hands of our research and robotics teams. You will be the single person turning capital into secured supply, ahead of demand and on the best terms in the market.
The role is commercial and strategic at its core: market intelligence, deal-making, and supply-chain execution across a tight and volatile accelerator market. It fits a leader who has negotiated large-scale compute agreements, knows how hyperscalers, neoclouds, and silicon vendors actually price and allocate, and can lock in capacity when everyone else is scrambling for it.
What You'll Own- Set and execute Luma's multi-year compute sourcing strategy: cloud vs. on-prem mix, hyperscaler vs. neocloud diversification, and long- vs. short-term commitment structuring.
- Own the full procurement lifecycle for accelerators and capacity - sourcing, RFPs, negotiation, contracting, and vendor management.
- Negotiate and secure the largest-scale capacity commitments and capital deployments, partnering with Finance on unit economics, pricing, and risk.
- Build and manage Luma's accelerator supply chain: allocation guarantees, delivery timelines, and contingency supply against shortages and demand spikes.
- Act as the principal executive interface and lead commercial negotiator to NVIDIA, AMD, hyperscalers, neoclouds, and frontier silicon vendors.
- Track the compute market continuously - pricing, availability, new entrants, custom silicon - and translate it into sourcing decisions that protect runway and unit economics.
What You Bring- Years of experience in large-scale technical procurement, infrastructure sourcing, or commercial partnerships, with a track record securing compute or capacity at a frontier AI lab, hyperscaler, or major autonomy program.
- Deep commercial fluency in the economics of large-scale training, GPU/accelerator pricing, allocation dynamics, and cloud contract structures.
- Proven record negotiating and operating multi-million-dollar (or larger) compute and infrastructure agreements.
Nice to Have- Relationships across NVIDIA, AMD, hyperscalers, and neoclouds, and a feel for how allocation decisions really get made.
- Experience sourcing capacity for training runs at the 100B-parameter or 100k-GPU-day scale.
- Enough technical fluency in cluster topology and interconnects (InfiniBand/RoCE) to evaluate what vendors are actually selling.
