NeuralStackSafeSuper AI announced a 2.8 billion dollar Series D on July 11 2026. Sequoia Capital and Andreessen Horowitz co-led the round with participation from Thrive Capital and existing investors. The round values the company at 35 billion dollars post-money.
Founded in 2023 the startup develops automated red-teaming and alignment evaluation platforms used by OpenAI Anthropic and Google DeepMind. Revenue reached 340 million dollars annualized in June 2026 up 280 percent year-over-year.
Funds will expand compute clusters in Texas and Ireland and hire 400 additional researchers. The company plans to open source portions of its evaluation harness by December 2026.
Earlier rounds included a 600 million dollar Series C in September 2025. Total funding now exceeds 4.1 billion dollars since inception.
Investors cited growing regulatory pressure on model safety as a key driver. Several governments have proposed mandatory third-party evaluations starting in 2027.
Why this matters
Safety tooling has become a distinct high-growth category separate from core model development. SafeSuper AI now sits alongside companies like Scale AI and Snorkel as essential infrastructure providers.
The valuation multiple of 103 times revenue reflects expectations that safety requirements will become table stakes for frontier model deployment. Smaller labs without access to such tools may face barriers to market entry.
Future rounds are expected to focus on government contracts as countries implement AI safety certification regimes. The outcome will determine whether safety remains a startup opportunity or becomes dominated by big tech internal teams.