Antithesis, a Northern Virginia startup pitching itself as infrastructure for never-down software, raised a $105 million Series A led by Jane Street, a bet that stress-testing distributed systems matters as much for blockchains as it does for high-speed trading.

The company’s platform uses deterministic simulation testing, running large-scale, production-like simulations to surface the kinds of edge cases that can blow up in live networks, Antithesis said in a press release Wednesday.

When a failure hits, Antithesis said it can replay the bug exactly, helping engineers isolate issues without the usual can’t reproduce limbo, a familiar pain point for crypto protocols where small glitches can cascade into chain instability.

Other investors in the funding round include Amplify Venture Partners, Spark Capital, Tamarack Global, First In Ventures, Teamworthy Ventures and Hyperion Capital, alongside individuals such as Patrick Collison, Dwarkesh Patel and Sholto Douglas, the company said.

Antithesis has leaned on crypto credibility, saying the Ethereum network used its simulations ahead of The Merge to model extreme conditions and catch vulnerabilities before the proof-of-stake transition.

The company also cited customers across finance, AI, blockchain and data infrastructure, and said revenue rose more than 12x over the past two years.

Antithesis said it will use the proceeds to expand engineering, increase automation, scale go-to-market globally and push distribution through channels including AWS Marketplace.

AI agents are now capable enough to identify exploitable weaknesses in smart contracts, and could already be used by attackers to weaponize those flaws, according to new research from Anthropic’s Fellows program.

Researchers said that as AI models get cheaper and more capable, automated hacking could spread from decentralized finance (DeFi) exploits to a wider range of software and critical infrastructure bugs.

Read more: Anthropic Research Shows AI Agents Are Closing In on Real DeFi Attack Capability