Compute for rent
Chips for rent, the memory and power behind them, the sites that come online, and what a GPU hour costs.
On the register
- 800 V DC rack power watching
- AMD Helios racks watching
- Accelerators the labs build for themselves watching
- Co-packaged optics switches watching
- Compute in Europe watching
- Desk-sized inference machines watching
- Energy per prompt disclosures watching
- Four-bit inference formats watching
- Gigawatt campuses coming online watching
- HBM4 supply watching
- Inference providers selling capacity watching
- Inference-only accelerators watching
- Ironwood TPUs for rent watching
- Labs committing to hyperscaler compute watching
- NVLink Fusion partners watching
- Neocloud power online watching
- Neural accelerators in Apple GPUs watching
- ROCm as a serving target watching
- UALink silicon watching
- Ultra Ethernet fabrics watching
- Vera Rubin systems watching
Everything filed here
Nothing yet in this section.