🔥 OpenAI Wins, Leopold's Bombshell 13F, and the Data Center Revolt

🏛️ OpenAI Dodges the Musk Lawsuit

A federal jury delivered a swift verdict in the high-profile case between Elon Musk and OpenAI CEO Sam Altman, finding Altman not liable for straying from the company's charitable mission. After just 90 minutes of deliberation, the jury unanimously ruled that Musk's claims were dismissed on timeliness grounds—he had simply waited too long to file the lawsuit.

The judge confirmed the verdict, and with that, Musk's lawsuit against OpenAI was officially over. No debate on the merits, no protracted courtroom drama—just a clean dismissal on statute of limitations. As Wired's Max Zeff summarized: "Musk loses the lawsuit against OpenAI."

For OpenAI, this represents a clear win and removes a significant legal distraction as the company continues its breakneck pace of development and fundraising.

📊 Leopold Aschenbrenner's 13F: The Most-Watched Filing in Years

Over the weekend, anticipation reached fever pitch for the release of Leopold Aschenbrenner's hedge fund 13F filing. Known for his aggressive thesis that frontier AI will continue improving at half an order of magnitude per year, Aschenbrenner has built a reputation for bold, AI-centric bets predicated on unprecedented demand for compute infrastructure.

When the filing finally dropped, it revealed massive put positions across the semiconductor sector, including $2 billion in puts on SMH, the VanEck Semiconductor ETF. The market interpreted this as a pointed bet—not a blanket skepticism of semiconductors, but a nuanced view on which companies are truly positioned to benefit from the next phase of AI infrastructure buildout.

Interestingly, the filing also showed that Aschenbrenner appears to be long Nvidia, despite ongoing debates about whether the company's moat remains intact amid rising competition. He also held positions in solar-related names, suggesting a belief that renewable energy timelines for powering AI data centers may be shorter than previously assumed—a notable shift from the fossil fuel-heavy narrative that dominated earlier AI infrastructure discussions.

Important Context: 13F filings are snapshots as of March 31, 2025. These positions are months old, and investors have no visibility into strike prices, expirations, hedge ratios, or whether options are part of broader strategies. Short positions and swaps are not disclosed. As one analyst put it: "Making investment decisions for assets based on data from months ago sounds like a good way to burn money."

The filing sparked widespread debate, with some calling it a masterclass in positioning and others cautioning against over-interpreting stale data. Either way, it marked a rare moment where a hedge fund's regulatory disclosure broke out of traditional finance circles and dominated tech and AI Twitter.

🏭 The Data Center Backlash: A Bipartisan Problem

AI data centers have become one of the most divisive infrastructure projects in America, drawing opposition from both the left and the right. The left worries about job displacement, the theft of creative work, and the erosion of human artistry. The right increasingly frames data centers as "surveillance centers"—tools for monitoring citizens—and views them as benefiting coastal elites at the expense of hollowed-out industrial towns.

This backlash has become so pervasive that even Eric Schmidt, former Google CEO, was loudly booed during a commencement speech at the University of Arizona when he mentioned artificial intelligence. The jeers persisted throughout his remarks, underscoring how deeply unpopular AI infrastructure has become among certain demographics.

🤠 The Kevin O'Leary Data Ranch

A flashpoint in this debate has been a proposed $100 billion data center project in Utah, championed by Shark Tank investor Kevin O'Leary. The project, called Stratos, would span over 40,000 acres and, at full capacity, consume 9 gigawatts of electricity—roughly double the entire current electricity usage of Utah.

Critics seized on O'Leary as a symbol of the problem. Known for his ostentatious style—he recently attended the Oscars wearing two luxury watches simultaneously—O'Leary became an easy target for those framing data centers as projects designed to enrich billionaires at the expense of local communities.

However, upon closer inspection, the Stratos project appears more defensible than headlines suggest. The data center is planned for Hansel Valley, an uninhabited desert area. It will generate its own power and water, using purchased agricultural water rights rather than tapping into local community resources. Proponents argue this is exactly the kind of remote, self-sufficient infrastructure that should alleviate concerns about grid strain and resource depletion.

Yet even this model has struggled to win public support. One viral video claimed the project would produce thermal energy equivalent to "dropping 23 atom bombs in Utah every single day." While this framing was quickly debunked—thermal load from power generation is not unique to data centers—the rhetoric illustrates how emotionally charged the debate has become.

Key Insight: The backlash isn't purely about the specifics of any given project. It reflects a broader frustration with AI itself, the concentration of wealth among tech elites, and a sense that transformative technologies are being built to communities rather than with them.

💰 Ben Thompson's Radical Solution: Just Pay People

Stratechery's Ben Thompson has proposed a novel—and deliberately crass—solution to the data center problem: start paying people directly.

Thompson argues that the promise of vague tax benefits funneled through local governments has failed to win over communities. Instead, he suggests data center operators should distribute annual cash payments to every resident of the towns where they build.

He ran the numbers on a 1.6 gigawatt data center that was proposed for the village of Deforest, which has around 11,500 residents. If the data center generates roughly $3 billion in annual operator revenue, paying every resident $10,000 per year would cost just 3.8% of annual revenue.

"I bet that proposal would have been approved," Thompson writes. "And I bet the operator could very easily pass on those costs to actual data center users."

This model treats local communities not as obstacles to be managed, but as stakeholders with a legitimate claim to the economic upside of hosting infrastructure they don't personally benefit from. It's a direct acknowledgment that AI is not a natural resource—users don't care where the data center is located, so there's little organic reason for locals to support construction unless they're compensated.

Whether this approach gains traction remains to be seen, but it represents a pragmatic shift in thinking about how to align incentives in infrastructure development.

🎓 The Consumer AI Problem

Underlying much of the backlash is a sense that AI has failed to deliver magical consumer experiences. In the early 2010s, cloud infrastructure enabled products like Uber, Yelp, and Groupon—services that felt immediately useful and delightful, even if they weren't world-changing.

By contrast, the current AI wave has produced little in the way of consumer delight. The focus has been on enterprise productivity, coding tools, and vibe-coded demos—leaving everyday users with chatbots that produce "clockable" writing and a nagging sense that the technology is overhyped.

As one observer noted: "Everyone is vibe coding 24/7, but the magical moments of consumer technology have been left behind."

This disconnect has real consequences. When Eric Schmidt tells college graduates to "help shape AI," students at a state university hear a different message: "The career paths I was hoping for are being automated." At Stanford, that pitch might land. At the University of Arizona, it gets booed.

🔮 What Comes Next

The AI infrastructure debate is far from settled. Nuclear power was once dismissed as too risky and unnecessary—"Electricity isn't that expensive," people said 50 years ago. Today, that decision is widely regarded as one of humanity's greatest mistakes, and it directly contributes to the challenges of powering AI data centers.

Ben Thompson argues that the tech industry needs to fix its messaging, control misinformation, and—most radically—start writing checks. Whether that happens, and whether it's enough to turn the tide, will shape the trajectory of AI infrastructure for years to come.

For now, the industry faces a sobering reality: building the future is hard when the present doesn't want it.