Market researcher IDC predicts that global businesses will invest $4.25 trillion in technology by 2026, with artificial intelligence (AI) being the main driver of this spending. According to a survey by venture capital firm Madrona, 74% of 150 enterprise IT professionals plan to increase their AI budgets over the next year, while the remaining 26% intend to keep their spending the same. Despite this optimism, fewer than half of AI pilot projects in enterprises move beyond testing and into full production. This is an improvement from last year, when a study by MIT found that 95% of AI projects failed to deliver a positive return on investment. Madrona’s report also highlights that even when AI tools are successfully deployed, enterprises often do not commit to them for the long term. Around 77% of companies re-evaluate their AI vendors every six months or continuously, leading to a "fast in, fast out" approach. This is different from traditional software-as-a-service (SaaS) models, where long-term contracts create a kind of resistance to change, known as a "moat of inertia." In the AI space, switching between vendors is easier, and the pressure to re-evaluate options is constant. This dynamic has implications for the financial health of AI startups. Many of these companies rely on annual recurring revenue (ARR), which is a measure of predictable income from ongoing subscriptions. In 2025, enterprise trial budgets fueled a surge in AI startups, and 2026 was expected to bring more long-term commitments from big companies. However, even after a startup’s product is adopted, securing stable revenue remains a challenge. One reason is that many AI startups have not yet developed clear pricing models for their products. Research by venture capital firm Andreessen Horowitz found that over half of technical AI buyers prefer to pay based on the work or outcomes delivered, rather than traditional usage-based metrics like the number of data tokens processed. This is different from the SaaS model, which often charges based on how much a service is used. For AI, pricing tied to recognizable outcomes—such as the number of reports generated, customer tickets resolved, or sales leads created—can help startups demonstrate their value to customers. This approach aligns the interests of both the startup and the enterprise, making the product more economically beneficial for both sides. Overall, AI seems to be encouraging more experimentation in the enterprise world, with companies more open than ever to trying new technologies. However, this also means that a single enterprise contract no longer guarantees long-term revenue for startups. Whether and when companies will shift back to more traditional, long-term purchasing habits remains uncertain.