Companies continue to increase their investment in AI, but this does not mean AI startups have secured stable, sustainable long-term revenue. Recent research shows that enterprise customers, while expanding their AI budgets, are reassessing vendors more frequently—the multi-year contract inertia common in the SaaS era is weakening in AI procurement.
Corporate budgets are still expanding.
A Madrona survey of 150 enterprise IT professionals found that 74% of respondents plan to increase their AI budgets over the next 12 months, while the remainder intend to maintain current spending levels. Meanwhile, fewer than half of AI pilot projects ultimately make it into production.
Market research firm IDC predicts that enterprise technology spending will reach $4.25 trillion by 2026, with the primary growth driven by AI. For startups, this means customers are still willing to try new products, and opportunities for pilot programs and procurement have not disappeared.
Renewal stability is declining
More importantly, even after deployment, companies may not retain the same vendor long-term. Madrona’s research shows that 77% of companies reassess their AI vendors every six months or on an ongoing basis.
This purchasing rhythm differs significantly from traditional enterprise SaaS. In the past, long-term contracts and high switching costs typically fostered strong customer loyalty; however, in the AI software space, businesses more easily replace tools and are more willing to continuously compare the performance and pricing of different products.
This means that even if AI startups move their products from pilot stages to full-scale adoption, the associated revenue may not be as reliably predictable as in the past. As a result, many companies' publicly reported ARR growth faces higher volatility risk.
Pay-for-performance is more popular.
A survey by venture capital firm Andreessen Horowitz of 50 technical AI buyers found that more than half of respondents prefer to pay for AI products based on the results achieved, rather than usage metrics such as token consumption.
Enterprises are more focused on how much work AI has actually accomplished, such as how many reports were processed, how many tickets were closed, and how many sales leads were generated. If pricing is directly tied to these outcomes, vendors can more easily demonstrate product value, making renewal and expansion more likely.