The AI narrative is changing before our very eyes. There is no longer a debate on whether it’s useful. In recent days, the focus has been on security, and AI threatening the human race. But there is another reality brewing. We are in a token war, and the implications for our economy are consequential.

In the last year, the biggest names in tech have gone on the largest borrowing spree in corporate history. Nvidia lined up a $500 billion debt facility with a consortium of Wall Street banks. Google raised $85 billion in new stock and in a shocking precedent, sold 100-year bonds. Oracle tapped markets for $45 billion combined in equity and debt. Amazon raised $25 billion in a six-part bond deal. Meta did its first bond offering in years, pulling in $30 billion. Intel sold $15 billion in stock just to keep pace with AI-related orders. It makes your brain hurt.

These are, by any measure, the most profitable companies on earth. So why do the most profitable companies need to borrow so much money? Because they are spending it.

These multi-trillion-dollar valuations rest on one assumption: that capital markets keep funding them.

America’s competitiveness lies in the balance. America’s advantage in this race is that private capital moves fast and takes real risk. Nobody in Beijing needs a bond prospectus to build a data center. China’s AI buildout is state-directed and doesn’t answer to quarterly earnings calls. If the U.S. capital markets get skittish, and they eventually will, our biggest strength, becomes our biggest vulnerability.

America’s AI buildout is running into strong opposition in the public square. It seems, no one wants datacenters in their backyard.

In the first three months of 2026 alone, 75 major data center projects worth more than $130 billion were delayed or canceled because local communities said no. On a single day in July, 142 separate protests hit 42 states. Peculiar, Missouri, a town of 3,000 people, killed a 500-acre data center campus outright under the banner “Don’t Dump Data on Peculiar.” Blackstone’s QTS abandoned what would have been the world’s largest data center campus in Virginia after a zoning technicality gave opponents an opening.

The complaints are consistent: water use (a large data center can drink as much as a city of 50,000 people), electricity prices up as much as 267% in some concentrated markets, noise, and a general sense that residents are absorbing the costs of an industry whose profits flow somewhere else. Gallup found 71% of Americans oppose a new data center in their community. That number is now showing up in governor’s races and House campaigns from Ohio to Pennsylvania.

Building data center bandwidth is what keeps the trains on the rails and may be the key to America’s economic and military advantage.

While AI’s finances get the headlines, AI’s role in cybersecurity is quietly becoming table stakes. Mid-market companies are behind.

Attackers are already using AI to automate reconnaissance and generate convincing phishing at a scale no IT team ever could, with lateral movement inside a breached network now compressed to under 30 seconds. The defensive side has responded in kind.

Leading platforms now run continuous behavioral monitoring across endpoints and identity systems, flagging anomalies in real time rather than waiting on a quarterly audit. Automated containment tools isolate a compromised device or account the instant it deviates from normal behavior, without waiting for a human to approve the response. Vendors like Fortinet have built generative AI directly into security operations centers, automating the triage work that used to eat an analyst’s entire shift.

The uncomfortable truth: most mid-market security programs were built for a 2023 threat model.

Moves worth making this quarter:

• Confirm your provider offers real-time behavioral monitoring, not just periodic scanning.
• Ask specifically what your system can contain automatically without waiting on a human.
• Budget and train as if the breach has already started.

In this fight, speed is the strategy. 

Ask ten people what excites them most about AI, and at least half will say some version of “it’s going to cure cancer.” While that is not quite reality yet, consider recent development of anti-cancer drugs.

Revolution Medicines just got FDA approval for daraxonrasib, Rasonque (daraxonrasib) a new treatment for pancreatic cancer, one of the deadliest diagnoses in oncology. In trials, it nearly doubled median survival compared to chemotherapy, 13.2 months versus 6.7. The drug came out of old-fashioned structural chemistry, years of bench science identifying a binding site nobody thought was possible.

Where AI showed up was after the fact. A company called Chemical.AI ran the approved molecule through its retrosynthesis engine and found a cheaper way to manufacture it, swapping the original 8-step process for a longer 17-step route built on far less expensive raw materials. Nobody asked them to. They did it as proof of concept for their own technology, and it worked.

There’s also a real possibility AI accelerates the regulatory side, faster data analysis, faster trial modeling, faster review cycles. That piece is still emerging, not proven, but it’s the next frontier worth watching.

Today on the show, twenty-five years after 9/11, America is still living with the consequences of the war on terror. It produced major tactical successes, at enormous human and financial cost. Fareed speaks with CNN national security analyst Peter Bergen, author of “All the Presidents’ Wars,” about the legacy of America’s war on terror.

Watch here

As CEOs plan for 2027, they continue to navigate the new realities that shape how companies compete, including global uncertainty and constant change. Rapid advancements in AI, shifting workforce dynamics, constant instability, and economic uncertainty are all underlying factors to consider when planning for the future.

Read here