Enterprise AI Contracts Are Less Secure, Research Shows

Enterprise companies are increasing their artificial intelligence budgets, but their commitments to AI vendors remain short-term, according to research from venture capital firm Madrona.

Madrona surveyed 150 enterprise IT professionals. Of those respondents, 74% said they plan to expand their AI budgets over the next 12 months, while the rest plan to keep spending steady. However, fewer than half of their AI pilots reach full production. Madrona described that result as an improvement over a 2025 MIT report, which found that 95% of enterprise AI projects had failed to deliver a return on investment.

Even after deployment, enterprises frequently reconsider their vendors. Some 77% re-evaluate their AI vendors every six months or on a rolling basis. Madrona said this creates a “fast in, fast out” pattern that differs from traditional enterprise SaaS, where multi-year contracts can create long-term stability.

Pricing is another challenge. Research from Andreessen Horowitz, based on a survey of 50 technical AI buyers, found that more than half preferred fees tied to work produced or other outcomes instead of usage measures such as tokens consumed.

The findings suggest that enterprise AI experimentation is opening opportunities for startups while making long-term revenue less secure, even after a product moves beyond its pilot phase.