Enterprise AI buys on speed

Companies are buying artificial intelligence models the way they buy database software now: on response time, price, and how cleanly they fit production systems. (futurumgroup.com) Futurum Group said on April 12 that 67% of organisations in its first-half 2026 survey already run generative artificial intelligence in production, and 75% expect to raise budgets in the next year. The firm surveyed 838 decision-makers and found OpenAI GPT at 61% enterprise adoption, ahead of Azure OpenAI at 50% and Google Gemini at 47%. (futurumgroup.com) Ramp said on April 11 that business artificial intelligence adoption crossed 50% for the first time in March, reaching 50.4% of businesses on its payments platform, up from 35% a year earlier. Ramp’s data showed Anthropic at 30.6% of businesses and OpenAI at 35.2%, shrinking the gap to 4.6 percentage points from 11 points in February. (ramp.com) That shift shows up in how companies deploy the tools. Futurum said customer service automation is the top generative artificial intelligence use case at 63%, a category where a slower answer can break a live workflow even if the model is more capable on paper. (futurumgroup.com) OpenAI’s own product changes point the same way. The company said on March 3 that GPT-5.3 Instant was built to respond faster and give more contextual web-search answers, and Futurum reported a new $100-a-month Pro tier aimed at power users and enterprise buyers. (openai.com) (futurumgroup.com) The money is moving beyond experiments. Menlo Ventures said enterprises spent $37 billion on generative artificial intelligence in 2025, up from $11.5 billion in 2024, with more than half of that spend going to applications rather than underlying model infrastructure. (menlovc.com) Menlo also said 76% of enterprise use cases are now bought rather than built internally, a sign that companies increasingly want packaged tools they can deploy quickly instead of custom systems they must maintain themselves. (menlovc.com) Vendors are still arguing over who has the stronger position. OpenAI told investors this month that its computing build-out gives it an advantage over Anthropic, even as spending and adoption data show Anthropic gaining ground with business customers. (bloomberg.com) (ramp.com) The current contest is less about a single best model than about which one clears a company’s budget, latency, and governance checks first. In enterprise artificial intelligence, speed has become a purchasing feature, not just a technical benchmark. (futurumgroup.com)