The Headline Number Is Real. The Victory Lap Is Not.

The March 2026 Ramp AI Index confirmed that 50.4% of U.S. enterprises now hold paid AI subscriptions — up from 46.8% in January. That sounds like a breakthrough.

It isn't.

A separate Conference Board survey of more than 250 HR leaders, released March 31, found that 60% of organizations are still in early-stage AI adoption — experimenting, not operating. Fewer than half have integrated AI into actual workflows. Only 11% report advanced enterprise-wide integration.

So roughly half of American companies are paying for AI. And most of them have nothing to show for it.

The Pentagon Guy Has Seen This Before

Drew Cukor knows what institutional AI failure looks like up close. He's the retired Marine colonel who stood up Project Maven — the Pentagon's landmark AI integration program — and later led AI transformation at JPMorgan Chase. Now president of TWG AI, he wrote in Fortune on May 11 that Corporate America is "making every mistake we almost made."

His diagnosis is blunt: companies are treating AI as an experiment to manage instead of an organizational transformation to own.

At Maven, that distinction almost killed the program. What saved it was senior leadership fighting for it personally, dismantling workflows instead of just layering AI on top of them, and measuring outcomes — what warfighters could actually DO — instead of tracking usage metrics.

Corporate America is doing the opposite on all three counts.

The Stanford Data Lands Like a Gut Punch

Cukor cites the Stanford 2026 AI Index, released in April, which ranked the U.S. 24th globally in AI adoption despite building the world's most powerful models. American adoption sits at 28.3%. Singapore is at 61%. The UAE is at 54%.

The country that built ChatGPT, Claude, and Gemini can't figure out how to use them.

Goldman Sachs put an even harder number on it: AI investment contributed "basically zero" to U.S. GDP growth last year. That's not fringe analysis — that's Goldman.

Our previous coverage flagged Uber's candid admission that AI ROI was missing in action. Now we have sector-wide data confirming Uber wasn't an outlier. It was the norm.

The AI Market Is Reshuffling Anyway

While companies struggle to use what they're buying, the vendor competition is getting brutal.

According to the Ramp AI Index, Anthropic's market share jumped from 24.4% to 30.6% in a single month — the largest monthly gain ever recorded in the index. The driver: Claude Code, a developer tool now running at a $2.5 billion revenue run-rate.

OpenAI, meanwhile, saw its share plateau at 35.2% and recorded a rare adoption dip. When businesses in 2026 are choosing a primary AI provider for the first time, they're picking Anthropic in roughly 70% of head-to-head matchups, per Ramp's data.

That's a seismic shift in a market that was supposed to be OpenAI's to lose.

What Mainstream Coverage Is Getting Wrong

Most business coverage is framing the 50% adoption milestone as proof the AI revolution has arrived. Adoption means someone bought a subscription. It does NOT mean the company transformed anything. The Conference Board data makes this clear — a majority of the organizations counted in that 50% figure are still in experimentation mode, with no measurable workflow integration and no real accountability for results.

Most coverage also overlooks the talent gap inside these companies. Per the Conference Board, 56% of HR leaders say AI fluency plays little or no role in promotions today. More than half of workers believe AI skills would help their career — but leadership isn't acting on it. You can't transform a company with AI if the people running it aren't being selected or rewarded for knowing how to use it.

The AI-Layoffs Myth Deserves a Reality Check Too

Fear-mongers on the left and the right have been screaming about AI mass layoffs. The Conference Board data says: only 6% of organizations cite AI as a primary reason for layoffs. Restructuring and financial pressure dominate. AI isn't firing people right now — it's just collecting a subscription fee.

What This Means for Regular People

If you're an employee wondering whether to upskill in AI, the answer is yes — but don't expect your company to reward you for it yet. Most organizations haven't figured out the strategy.

If you're a taxpayer watching government tout AI productivity gains, look for the proof. Goldman Sachs already told you the GDP number: zero.

If you're an investor in AI infrastructure plays banking on enterprise transformation revenue, the gap between adoption and actual deployment is your biggest risk right now.

Cukor's core lesson from Maven: the technology was never the problem. The leadership discipline to actually use it was. Corporate America has the technology. It is failing the discipline test — badly.