Microsoft drops AI sales targets in half after salespeople miss their quotas
Report: Microsoft declared “the era of AI agents” in May, but enterprise customers aren’t buying.
Microsoft has lowered sales growth targets for its AI agent products after many salespeople missed their quotas in the fiscal year ending in June, according to a report Wednesday from The Information. The adjustment is reportedly unusual for Microsoft, and it comes after the company missed a number of ambitious sales goals for its AI offerings.
AI agents are specialized implementations of AI language models designed to perform multistep tasks autonomously rather than simply responding to single prompts. So-called “agentic” features have been central to Microsoft’s 2025 sales pitch: At its Build conference in May, the company declared that it has entered “the era of AI agents.”
The company has promised customers that agents could automate complex tasks, such as generating dashboards from sales data or writing customer reports. At its Ignite conference in November, Microsoft announced new features like Word, Excel, and PowerPoint agents in Microsoft 365 Copilot, along with tools for building and deploying agents through Azure AI Foundry and Copilot Studio. But as the year draws to a close, that promise has proven harder to deliver than the company expected.
According to The Information, one US Azure sales unit set quotas for salespeople to increase customer spending on a product called Foundry, which helps customers develop AI applications, by 50 percent. Less than a fifth of salespeople in that unit met their Foundry sales growth targets. In July, Microsoft lowered those targets to roughly 25 percent growth for the current fiscal year. In another US Azure unit, most salespeople failed to meet an earlier quota to double Foundry sales, and Microsoft cut their quotas to 50 percent for the current fiscal year.
The sales figures suggest enterprises aren’t yet willing to pay premium prices for these AI agent tools. And Microsoft’s Copilot itself has faced a brand preference challenge: Earlier this year, Bloomberg reported that Microsoft salespeople were having trouble selling Copilot to enterprises because many employees prefer ChatGPT instead. The drugmaker Amgen reportedly bought Copilot software for 20,000 staffers only for them to ignore it in favor of OpenAI’s chatbot.
A Microsoft spokesperson declined to comment on the changes in sales quotas when asked by The Information. But behind these withering sales figures may lie a deeper, more fundamental issue: AI agent technology likely isn’t ready for the kind of high-stakes autonomous business work Microsoft is promising.
The gap between promise and reality
The concepts behind agentic AI systems emerged shortly after the release of OpenAI’s GPT-4 in 2023. They typically involve spinning off “worker tasks” to AI models running in parallel with a supervising AI model, and incorporate techniques to evaluate and act on their own results. Over the past few years, companies like Anthropic, Google, and OpenAI have refined those early approaches into far more useful products for tasks like software development, but they are still prone to errors.
At the heart of the problem is the tendency for AI language models to confabulate, which means they may confidently generate a false output that is stated as being factual. While confabulation issues have reduced over time with more recent AI models, as we’ve seen through recent studies, the simulated reasoning techniques behind the current slate of agentic AI assistants on the market can still make catastrophic mistakes and run with them, making them unreliable for the kinds of hands-off autonomous work companies like Microsoft are promising.
While looping agentic systems are better at catching their own mistakes than running a single AI model alone, they still inherit the fundamental pattern-matching limitations of the underlying AI models, particularly when facing novel problems outside their training distribution. So if an agent isn’t properly trained to perform a task or encounters a unique scenario, it could easily draw the wrong inference and make costly mistakes for a business.
The “brittleness” of current AI agents is why the concept of artificial general intelligence, or AGI, is so appealing to those in the AI industry. In AI, “general intelligence” typically implies an AI model that can learn or perform novel tasks without having to specifically be shown thousands or millions of examples of it beforehand. Although AGI is a nebulous term that is difficult to define in practice, if such a general AI system were ever developed, it would hypothetically make for a far more competent agentic worker than what AI companies offer today.
Despite these struggles, Microsoft continues to spend heavily on AI infrastructure. The company reported capital expenditures of $34.9 billion for its fiscal first quarter ending in October, a record, and warned that spending would rise further. The Information notes that much of Microsoft’s AI revenue comes from AI companies themselves renting cloud infrastructure rather than from traditional enterprises adopting AI tools for their own operations.
For now, as all eyes focus on a potential bubble in the AI market, Microsoft seems to be building infrastructure for a revolution that many enterprises haven’t yet signed up for.
Benj Edwards is Ars Technica’s Senior AI Reporter and founder of the site’s dedicated AI beat in 2022. He’s also a tech historian with almost two decades of experience. In his free time, he writes and records music, collects vintage computers, and enjoys nature. He lives in Raleigh, NC.
