As a Builder of the Pentagon’s AI Overhaul, I See Corporate America Repeating Our Near-Mistakes

(SeaPRwire) –   When we were establishing Project Maven, the Defense Department’s initiative to integrate AI into some of the world’s most complex and high-stakes operational systems, the skeptics within the Pentagon were not mistaken in their doubts. The department had a lengthy and costly history of technology projects that consistently arrived late, exceeded budgets, and delivered insufficient results. There was no clear reason to believe AI would be any different.

What set Maven apart wasn’t the underlying technology. Instead, it was the decision to treat AI not as a limited experiment requiring management oversight, but as a broad organizational transformation demanding full ownership. Senior leaders championed it both personally and through bureaucratic channels. Existing workflows weren’t just modified—they were replaced. Success was measured solely by outcomes: what warfighters could actually accomplish that they couldn’t before. That level of discipline made the project successful.

I share this story because Stanford’s 2026 AI Index, published in April, confirmed what corporate America has been quietly acknowledging in earnings calls all spring: the United States, which builds the world’s most powerful AI models, ranks 24th globally in how effectively it uses them. Current U.S. adoption stands at 28.3 percent. Singapore leads with 61 percent, while the UAE sits at 54 percent. Goldman Sachs recently observed that AI investments contributed “basically zero” to U.S. GDP growth last year. America isn’t falling behind due to inferior models or semiconductors. It’s lagging for the same reasons the Pentagon nearly lost momentum on Maven—and the solution is the same one that ultimately saved it.

The U.S. excels at creating cutting-edge artificial intelligence but struggles significantly when it comes to deploying it effectively. For now, China’s advantage doesn’t stem from superior technology—it stems from superior integration. And in the competition that truly matters, integration will determine victory.

China’s “AI Plus” strategy explicitly focuses on embedding AI across critical sectors including manufacturing, logistics, scientific research, healthcare, education, and government operations. In manufacturing specifically, the approach moves beyond generic assistants toward sector-specific models, industrial data sets, intelligent agents, and large-scale workflow integration. China isn’t treating AI as something to develop—it’s treating it as essential infrastructure. They aren’t debating whether AI can be controlled or contained; they are actively deploying it.

History abounds with examples of nations and companies that led in innovation but failed in implementation. We often rest comfortably on our national strengths in AI models and chip design, citing leaderboard rankings and benchmark scores. Yet these achievements, while impressive, don’t ultimately determine who wins the real-world race.

This explains why much of the current corporate AI discourse feels fundamentally misguided. Executives focus on spending and use cases. Procurement is straightforward, and use cases appear promising—but they’re the wrong metric. Organizational restructuring is seen as threatening. It forces leaders to confront uncomfortable questions they’ve long avoided: Which decisions should be automated? Which manual reviews can be eliminated? Which outdated workflows must disappear?

Most corporations don’t want honest answers to those questions. They seek the appearance of innovation without enduring the disruption of change.

Building an organization around AI is inherently difficult—which is precisely why American businesses are failing to meet this challenge. The companies that thrive over the next decade will be those willing to completely rebuild themselves, dismantling legacy processes, outdated organizational charts, and long-held assumptions about how work gets done. They will emerge as genuinely AI-native organizations—not merely augmented by AI. This distinction between augmentation and native integration is the difference between staying competitive and becoming obsolete.

This isn’t just a theoretical concern. It’s an emerging crisis with a distinct shape: a white-collar reckoning likely more disruptive than the blue-collar offshoring wave of the early 1970s, but far faster and less forgiving. If American companies fail to act decisively, they will face increasing pressure from Eastern competitors who are leaner, more agile, and free from the institutional inertia that is rapidly becoming corporate America’s greatest liability.

Maven was never intended as a pilot program. It was conceived as a comprehensive organizational transformation—one that proved a premise many doubted: that commercial software and AI could be successfully embedded into the workflows of the world’s largest bureaucracy, handling some of its most complex tasks, and delivering results at a scale legacy systems could never achieve.

The Defense Department is hardly renowned for its procurement successes. When it comes to software development, it’s known for producing expensive, delayed solutions that underperform. Yet, on Maven, the department got one crucial thing right: rather than launching a narrow technology experiment, it approached AI as an institutional transformation requiring top-down executive ownership from day one. That lesson is vital for corporate America to embrace.

Today, most large companies run isolated AI experiments. Far fewer undertake genuine AI transformations. The gap isn’t technological—it’s organizational. Maven’s experience reveals three critical failures that corporate America must address:

First, scaling AI requires executive ownership, not delegation. At the Pentagon, Maven succeeded because senior leaders owned it personally and advocated for it within bureaucratic structures. In too many corporations, AI strategy is handed off to a chief AI officer or an innovation lab—organizational constructs designed to signal progress while leaving the existing power structure untouched. This approach traps initiatives in perpetual pilot purgatory.

Second, AI demands the dismantling of legacy processes, not superficial layering atop them. Companies that simply bolt AI onto existing workflows achieve only marginal efficiency gains and call it transformation. If your organizational chart, approval chains, and operating rhythm remain unchanged after launching an AI initiative, you haven’t transformed anything.

Third, AI requires outcome-based accountability. Maven was evaluated strictly by operational impact: what new capabilities it enabled warfighters to achieve. Too many corporate AI programs are judged by activity metrics: number of models trained, proofs of concept completed, or partnerships announced. Activity without tangible outcomes amounts to little more than performative theater.

The response to China’s AI advancement isn’t better models or superior processors. It’s superior execution. America can still lead the AI era—but only when its institutions look inward and prove they possess the capacity for fundamental change.