Shares of Microsoft are down Wednesday following reports that the company has slashed growth targets for its AI software. The Information alleges that enterprise adoption of Microsoft Foundry, the company’s platform for building autonomous agents, is lagging significantly behind internal projections.
Swiftly rejecting the claims, a Microsoft spokesperson stated the company has not lowered “sales quotas.” Despite the denial, the news crystallized investor fears that revenue from complex AI tools is failing to match the industry’s $400 billion infrastructure spend.
Compounding the narrative of friction, the report lands just days after Microsoft warned of “novel security risks” in agentic AI and saw the Zig Software Foundation quit GitHub over the company’s “AI obsession.”
Promo
The Disconnect: Sales Miss vs. Semantic Defense
Market reaction was swift, with Microsoft shares dipping over 2% immediately following the sales miss report from The Information. At the heart of the sell-off is the allegation that internal growth targets were slashed from an ambitious 100% (doubling) to approximately 50% or lower.
Performance metrics from specific U.S. Azure units reportedly reveal the extent of the struggle. In one division, fewer than 20% of salespeople managed to hit a 50% growth target. Another unit was forced to slash its ambitious goal of doubling sales in half after the vast majority of the team failed to meet the original benchmark.
Microsoft issued a rapid, specific denial to CNBC. A spokesperson explicitly stated that “Aggregate sales quotas for AI products have not been lowered, as we informed them prior to publication.”
Investors, however, are reading between the lines. A company can miss high-level growth targets while leaving individual quotas unchanged, or simply watch staff fail to hit them.
Suggesting a narrative battle over internal metrics rather than a refutation of the underlying demand softness, the company argues the report is flawed, stating: “The news outlet inaccurately combined the concepts of growth and quotas.”
The Technical Friction: Why Agents Aren’t Selling
Crucially, the sales weakness is not in general SaaS like Copilot for Microsoft 365, but specifically in Microsoft Foundry, formerly Azure AI Studio.
This platform is designed for building custom AI agents, which are the industry’s current north star, promising autonomous workflows rather than simple chat-based assistance.
However, the reality of deploying these agents is hitting a wall of security and reliability concerns. Just weeks prior, Microsoft quietly updated support documentation to warn of “novel security risks” inherent to these tools.
Updated support documentation explicitly defines the mechanics of the threat:
“Agentic AI applications introduce novel security risks, such as cross-prompt injection (XPIA), where malicious content embedded in UI elements or documents can override agent instructions, leading to unintended actions like data exfiltration or malware installation.”
For enterprises, this creates a “liability loop.” The more autonomous the agent, the higher the risk of it executing a malicious command. Security researchers have compared this vulnerability to the “Office Macros” era, where functionality created a massive attack surface.
Kevin Beaumont, a prominent security researcher, characterized the new feature as “Macros on Marvel superhero crack.” Beyond security, basic integration remains a hurdle. The report cited private equity firm Carlyle struggling to get tools to reliably connect disparate data sources.
The Cultural Friction: The Developer Revolt
While enterprise buyers hesitate, the developer community, critical for building the “Foundry” ecosystem, is showing signs of revolt. The Zig Software Foundation publicly announced Zig’s departure from GitHub to Codeberg this week.
Framed as a high-profile rejection of Microsoft’s strategy, Zig leadership claims the pivot has degraded core service quality. Specific technical grievances include a “safe sleep” bug in GitHub Actions that went unfixed for months, clogging CI/CD pipelines.
Andrew Kelley, President of the Zig Software Foundation, argued that “Priorities and the engineering culture have rotted, leaving users inflicted with some kind of bloated, buggy JavaScript framework in the name of progress.”
To make this point, the foundation is taking a significant financial hit, walking away from approximately $170,000 in annual revenue from GitHub Sponsors.
Aligning with broader developer sentiment, this exodus highlights fears that prioritizing AI compute is cannibalizing the reliability of basic hosting tiers.
The Macro Stakes: The $400 Billion Gamble
These specific friction points, sales misses, security warnings, developer churn, are converging to challenge the broader AI economic thesis. Currently, the industry is locked in a capital expenditure cycle, spending historic sums on infrastructure.
Lisa Shallet, Chief Investment Officer at Morgan Stanley Wealth Management, noted that “Hyperscaler capex on data center and related items has risen fourfold and is nearing $400 billion annually.”
Consequently, market anxiety has shifted. In October, the fear was accelerated capital spending; in December, the fear is earning too little. The gap between the massive infrastructure build-out and the friction-filled enterprise adoption is widening.
If “Agentic AI” cannot be safely or reliably deployed, the high-margin revenue layer required to justify the GPU spend evaporates. This dynamic mirrors the dot-com fiber glut, where infrastructure was built years before the applications arrived to utilize it.