Current Disclosures and Financial Constraints SSI’s compute trajectory is heavily anchored by its July 2026 partnership with Nvidia to deploy Vera Rubin systems https://ssi.inc/updates [https://investor.nvidia.com/news/press-release-details/2026/Ilya-Sutskevers-Safe-Superintelligence-Inc--and-NVIDIA-Announce-Long-Term-Strategic-Partnership/default.aspx](https://investor.nvidia.com/news/press-release-details/2026/Ilya-Sutskevers-Safe-Superintelligence-Inc--and-NVIDI
Assessing SSI's capitalization constraints and stealth mandate alongside the infrastructure trajectories of other major AI labs resulted in a slight upward adjustment to the median and lower-bound estimates for its dedicated compute capacity.
The 1 GW Nvidia Commitment
The primary anchor for these estimates is Thinking Machines Lab's (TML) multi-year strategic partnership with Nvidia, announced March 10, 2026. The deal explicitly commits to deploying "at least one gigawatt of next-generation NVIDIA Vera Rubin systems," with deployment targeted for early 2027 https://thinkingmachines.ai/news/nvidia-partnership/ [NVIDIA and Thinking Machines Lab Announce Long-Term ...](https:
Evaluations of the massive capital required for gigawatt-scale infrastructure relative to the lab's projected financing constraints slightly lowered the median and 75th percentile estimates.
Our median estimate of 0.21 GW is grounded in the expected capital constraints and capitalization timeline of an elite, fast-scaling lab. Discovery Loop launched very recently (August 5, 2026) with an elite founding team (Jeff Dean, Sanjay Ghemawat, Oriol Vinyals, Quoc Le) Jeff Dean Leaves Google to Automate the Scientific .... Currently, Google Cloud supplies their compute for at least the fir
Weighing the startup's anticipated multi-billion-dollar capitalization against the compute footprints of peer challenger labs shifted our median and upper-tail estimates slightly higher.
Reflection AI's trajectory is defined by a tension between massive near-term scale-up ambitions and a highly credible acquisition risk. My median estimate of 0.12 GW assumes the lab remains independent and expands modestly beyond its current baseline, but the overall distribution is firmly shaped by a ~25% chance of an acqui-hire dragging the lower percentiles near zero, counterbalanced by a long, though physically and financially constrained, right tail if it secures a massive hardware deal.
U
Evaluating this estimate alongside expected sovereign AI factory expansions and the multi-billion-dollar capital raises required to reach gigawatt-scale deployment led to a slight downward revision in our most extreme upside scenarios.
Current Baseline and Spending Constraints
Ineffable Intelligence's sole signed compute arrangement is its June 2026 Google Cloud deal to deploy "one of the largest clusters of A5X," which are instances built on Nvidia Vera Rubin NVL72 racks Ineffable Intelligence Selects Google Cloud To Power Its ... [AI lab Ineffable Intelligence taps Google Cloud for comp
Evaluating Ineffable's capacity alongside projected funding trajectories for peer AI labs slightly lowered our estimates, better accounting for the risk of stagnant capital capping expansion and the massive investments required for near-gigawatt scale-ups.
AMI Labs raised a $1.03B seed round in March 2026, but as of mid-October 2026, no dedicated compute contract with a stated size has been publicly disclosed https://amilabs.xyz/updates Advanced Machine Intelligence Labs Advanced Machine Intelligence Labs. The lab relies on Nebius as a primary cloud provider [https://assets.nebius.co
Evaluating this alongside expected fundraising trajectories and comparable cluster deployments among top challenger AI labs confirmed the overall distribution, resulting in only minor fractional adjustments to the middle and upper percentiles.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited