Like millions of other sports fans, I spent the weekend celebrating the sacred rites of March Madness. In keeping with tradition, that meant losing my mind over dramatic moments like this, choosing a doomed underdog in whom to over-invest emotionally (I’ll miss you, LeBron Frames), and eating nachos for dinner on purpose.

What I didn’t do was bet on the games—and in the era of the smartphone sports-betting app, that makes me an outlier. The American Gaming Association estimates that hoops fans will wager $3.3 billion this year on the men’s and women’s NCAA basketball tournaments, up 54% since 2023. And that’s just counting legal, state-regulated gambling; throw in countless office pools, some 36 million online brackets, and the legal gray area of prediction markets like Kalshi and Polymarket, and the total easily climbs above $10 billion.

The three twenty-something finance guys sweating out Michigan’s first-round win on Thursday night (trust me, it was close for a while) at the table next to mine in a Manhattan Irish pub clearly had skin in the game. They were also glued to their phones. Sportsbook apps like the ones they were using—think DraftKings, or Flutter Entertainment’s FanDuel—have made betting both easy and “sticky,” using algorithm-driven, personalized pitches to push more wagering options out to their power users. Those apps represent an unusually potent combination of tech’s tools of enticement: the dopamine hits of online engagement, the wave of beguiling customized buying opportunities, and of course, the timeless exhilaration of the occasional win.

Like every great vice, betting on sports is a thrill until it becomes a problem. Studies suggest gambling addiction numbers have risen steadily as online betting has gotten more popular, and powerful tech is as always a double-edged sword, multiplying both the excitement and the perils.

As for me, I’ve got an addictive personality and an affinity for lost causes, so I’m staying off the apps and saving my money for next weekend’s nachos.

Matt Heimer
@MatthewHeimer
matt.heimer@fortune.com

Want to send thoughts or suggestions to Fortune Tech? Drop a line here.

Was this newsletter forwarded to you? Subscribe here!

Anduril's new contract is a $20 billion turning point for the Pentagon

Tech defense startups doing business with the U.S. military could someday look back at March 2026 as the month their relationships moved into the serious phase. Instead of dabbling in limited pilot projects with the startups, the Pentagon is starting to place big bets on a select few of these companies, writing them into core missions with the kind of fixed-priced deals that have long been standard among massive, established defense contractors like Lockheed Martin or Boeing.

Earlier this month, the U.S. Army announced an enormous deal with Anduril—a five‑ to ten‑year enterprise contract with a ceiling of up to $20 billion—that consolidates roughly 120 to 130 existing orders they already have, for both software and hardware, under one umbrella and creates a one‑stop vehicle to cut future deals much faster. The Army already inked a brand-new $87 million contract with Anduril earlier this week, as the first task order under that agreement. 

For venture-funded defense tech startups, which make everything from AI-powered drones to advanced threat detection systems, Anduril’s long-term contract sets a new bar that reflects how the young industry has evolved in the past few years—and opens the door to new opportunities and risks.—Jessica Mathews

Cursor's rapid rise and uncertain future

It’s a story distinctly of the AI era: Cursor, the buzzy AI coding startup, is four years old—and it already has an innovator’s dilemma.

Cursor’s annualized revenue crossed $2 billion in February and has marched well past that since, investors say. As of this weekend, the company was reportedly raising a new funding round that would value it at $50 billion.

But Cursor has a problem, and that problem is called Claude Code, a competitor launched by Anthropic barely a year ago that helped unleash a revolution in coding via agentic AI. Some say Claude Code, bolstered by Anthropic’s $380 billion largesse, could soon unseat Cursor altogether. Reports have surfaced of customers moving off Cursor; one Cursor investor told Fortune that several startups in his portfolio are decoupling from it. 

In a new feature, Fortune looks at Cursor's rapid rise and its current conundrum, and talks with CEO Michael Truell and President Oskar Schulz about their high-stakes effort to adapt on the fly.—Allie Garfinkle

Trump's White House gives Congress an AI road map

The White House has released a federal AI policy framework it’s asking Congress to turn into law. As expected, the blueprint favors light-touch federal oversight over the current state-level rules, urging Congress to preempt state AI laws that the administration has said is burdensome to innovation.

The federal policy carves out exceptions for child safety protections, including bans on AI-generated child sexual abuse material, and asks lawmakers to introduce age-gating requirements for AI models likely to be accessed by minors.

The framework also calls for legislation encouraging AI skills training, data collection on AI-driven job displacement, and codification of a ratepayer protection pledge—already signed by Amazon, Google, and OpenAI—requiring tech companies to supply or pay for the electricity used by the data centers they operate.—Beatrice Nolan

More tech

—Elon Musk announces a Tesla/SpaceX/xAI chip collaboration. It's called Terafab and sounds ginormous.

—OpenAI to double its headcount this year. The Financial Times says the AI giant is aiming for a staff of 8,000.

—How Mark Zuckerberg is using an AI agent at work. "We're going to get a lot more done," he says.

—Amazon notches its highest-grossing movie debut to date. 'Project Hail Mary' connects.

—Why 'metacognition' is the key to smarter use of AI. It also sounds fun; maybe you're already doing it?

—A Florida man let ChatGPT sell his home. Bad news for real-estate agents: It went really well.