Safe Superintelligence Inc. (SSI), founded in mid-2024 by Ilya Sutskever, has maintained a strict “straight-shot” focus on aligned superintelligence without releasing any models, papers, or public benchmarks in its first 25 months. The July 2026 NVIDIA partnership—providing an order-of-magnitude increase in Vera Rubin compute and several billion dollars in investment—signals that internal research has reached a stage worth scaling, yet the announcement contained no roadmap, capability claims, or release commitments. Investor speculation about an August 2026 model launch remains unconfirmed and secondhand, consistent with SSI’s deliberate avoidance of incremental products. Achieving and publicly announcing superintelligence by year-end would require unprecedented technical acceleration in roughly four months, far beyond typical large-model development cycles or the lab’s stated safety-first priorities. These factors underpin the market’s strong 88% implied probability on “No.”