Glean hit $200M ARR on February 17, 2026—doubling from $100M in just nine months. But the real story isn’t Glean’s growth. It’s whether neutral middleware can survive when Microsoft bundles AI into every tool your company already pays for.
The $465B infrastructure buildout is happening right now, and most enterprises are choosing based on sales pitches, not performance data. Glean’s velocity proves companies will pay to avoid vendor lock-in—but the math on whether middleware actually delivers that freedom remains unproven.
The infrastructure war has already picked winners—middleware wasn’t invited
Glean reached a $7.2B valuation without burning billions on compute like the frontier labs everyone watches. The bet here is fundamentally different: integration efficiency over model performance. CEO Arvind Jain is selling the promise that enterprises can plug into any AI—GPT-5, Claude Opus 4.5, Gemini 2.0—without marrying Microsoft or Google’s ecosystem.
The market believes him. Enterprise AI software hit $60B in 2026 with 50% year-over-year growth, according to TLDL AI Market Map data. Infrastructure spending is projected to balloon from $90B this year to $465B by 2033—a 24% compound annual growth rate that makes every layer of the stack a potential monopoly.
But the $465B infrastructure war is forcing major players to choose sides. And most are betting on bundled ecosystems, not neutral middleware. Microsoft Copilot and Google Duet AI offer zero-friction deployment because they’re already woven into Word, Gmail, Teams—the tools enterprises can’t quit. Glean promises freedom. The giants promise convenience. Enterprises have to pick before they know which one matters.
The 85% failure rate doesn’t care about your vendor strategy
Here’s the honest limitation: 85-95% of enterprise AI projects still fail, and 80% see zero ROI. An MIT study from mid-2025 found that 90% of generative AI pilots never make it to production. The middleware promise—”integrate everything without lock-in”—assumes the problem is vendor choice, not implementation complexity.
But 82% of firms can’t prove their AI works, regardless of who built it.
And bundled AI isn’t solving this either. Microsoft projects $11B in Copilot revenue for 2026 while claiming 90% Fortune 500 adoption—yet workers across industries report AI is creating more work than saving it. Gartner data shows 72% of Copilot users struggle with daily workflows. Integration friction exists whether you’re locked into one vendor or juggling three.
The middleware thesis rests on a promise—”don’t get locked in”—that enterprises can’t validate until years into deployment. No Fortune 500 companies have publicly disclosed rollback cases. No comparative productivity data exists. Glean’s growth proves demand. It doesn’t prove the solution works.
The real cost is the governance nightmare you can’t see yet
Middleware promises to prevent vendor lock-in. But enterprises risk data silos, integration headaches, and compliance costs if Glean—or any neutral layer—can’t fully neutralize the gravitational pull of Microsoft and Google ecosystems.
IT teams face a choice: bet on middleware freedom that requires custom integration work, or accept bundled AI convenience that might lock them in for a decade. No one has enough deployment data to know which risk is worse. There’s no pricing transparency—no public per-employee costs, no contract minimums, no documented governance failures from the past three months.
Glean’s $200M ARR proves enterprises will pay to avoid lock-in. The 85% failure rate proves most of them can’t tell if they’re buying freedom or just renting a different kind of dependency. The $465B infrastructure buildout is happening either way—and most companies are choosing based on promises, not proof.
