At the Republican midterm convention in Dallas this week, Vice President JD Vance dropped a number that should stop any taxpayer in their tracks: $250 billion in fraud identified since Vance began helming the administration’s anti-fraud task force in March.
This is not abstract “government waste.” It is real money, stolen from programs for sick kids, disabled veterans, and struggling families, and redirected into the pockets of criminals running fraud rings.
Vance’s investigation was enabled by AI and the data centers underpinning it. The Task Force to Eliminate Fraud, working closely with the Centers for Medicare & Medicaid Services (CMS), deployed AI systems with capabilities the federal bureaucracy previously could not implement at scale: spot patterns in real time.
Before this technology existed, catching a fraudulent provider meant a human being noticing something wrong, opening an investigation, and manually working through the red tape to shut it down—a process that could take months or years while the fraud kept bleeding taxpayers dry, as Fox News reported when the task force first rolled out its AI platform.
Now, AI models scan claims as they come in, flag anomalies, and can block or suspend suspicious billing automatically. That’s how CMS was able to identify and suspend 70 hospice and home health providers in Los Angeles within a single week of being flagged as high-risk—a number that later grew to 447 hospices and 23 home health agencies tied to more than $600 million in suspected fraud, according to reporting from the Washington Examiner.
The law firm Duane Morris, which tracks healthcare regulatory action for its clients, confirmed that CMS suspended Medicare payments to these providers based in part on the task force’s AI-assisted review of billing data. That’s how investigators zeroed in on the Medicaid fraud draining a Minnesota autism program, which Vance referenced directly in his convention remarks. This isn’t hypothetical. AI is doing the unglamorous, tedious work of catching thieves faster than any team of humans ever could.
And this is exactly why the conversation happening in communities across the country about data centers—the ones being built to power this new generation of AI—deserves greater nuance.
There’s a lot of noise right now about data centers: worries about water use, power grids, and “AI hype.” Some of that skepticism is fair, and any responsible buildout should account for local resources and grid capacity. But too much of the conversation ignores what these facilities make possible. The same infrastructure critics dismiss as a nuisance is the backbone of the systems now catching fraud that has bled Medicare and Medicaid for decades.
America cannot run AI fraud-detection models scanning millions of claims a day without the data centers to power them. New server racks provide the government the tools to protect taxpayer dollars. Criminals use high-tech tools to bilk taxpayers; it’s only fair that government (on behalf of taxpayers) fights back fire with fire.
Even outlets skeptical of the administration’s broader efficiency push have acknowledged the premise here. As MIT Technology Review noted, fraud is measurable, and machine-learning models applied to government data hold real promise for catching it faster than manual review ever could.
AI-driven detection is catching real fraud, faster than the old system ever could, and it’s happening because the data center infrastructure exists to run it. Conservatives have spent decades wanting the government to run more like a business—efficient, accountable, allergic to waste. That’s why President Trump, a businessman, has been so vocally supportive of data centers.
Vance’s task force, powered by AI and the data centers that make AI possible, is the first real evidence in a long time that it’s achievable.
The next time someone in your community complains about a data center going up, tell them: That’s not just a building full of servers. That might be what finally catches the fraud no human could find.