Andreessen Horowitz invests in Vals
Funding Provisional 90% confidence first seen
Andreessen Horowitz led Vals’s $40 million Series A, as reported by TechCrunch. The funding is described as enabling Vals to replace outdated AI evaluation methods with industry-task and impact-focused testing, while expanding staffing from 8 to 25 and relocating. The deal matters because improved, harder-to-game benchmarks can help AI companies validate model capabilities and measure real-world impact.
The deal
Vals $40 million Series A · announced 19 Sep 2026
Investors Andreessen Horowitz
Deal terms as reported in the coverage below.
Decision brief
- What changed
- Vals, an AI benchmarking startup founded in 2024, raised a $40 million Series A led by Andreessen Horowitz, according to TechCrunch. The company says it will use the funding to expand from 8 to 25 employees, relocate, and continue building industry-task and impact-focused AI evaluation services, including work for federal agencies.
- Why it matters
- For leaders deploying or building AI systems, this signals more investment flowing into evaluation tools that aim to measure model performance on real-world tasks rather than easier-to-game benchmark tests. Better benchmarking could improve vendor selection, model validation, and claims about business impact, especially in regulated or high-stakes environments. The federal-agency angle also suggests potential demand for more rigorous evaluation standards across sensitive use cases.
- Evidence
- The coverage is based on a single TechCrunch report stating that Andreessen Horowitz led Vals’s $40 million Series A and that Vals is focused on replacing outdated AI evaluation methods with industry-task and impact-focused testing. The same report says the company plans to grow headcount, relocate, and evaluate models for federal agencies; no additional independent reporting is provided here.
- What remains uncertain
- It is not yet clear how differentiated or widely adopted Vals’s benchmarking approach will become versus other evaluation providers, or whether its methods will prove materially harder to game in practice. The commercial traction, specific federal-agency engagements, and measurable customer outcomes are not established in the provided coverage.
- Monitor next
- Watch for named enterprise or government customers and published case studies showing Vals’s benchmarks changing model purchasing, deployment, or compliance decisions.
Analytical support, not advice — assumptions and open questions stated above.