One Disney Employee Called Claude 51,000 Times a Day — And Nobody Asked Permission

One Disney Employee Called Claude 51,000 Times a Day — And Nobody Asked Permission

Someone at Disney didn't wait for an AI strategy memo. They just started calling Anthropic's Claude — the AI assistant built by the $61 billion startup — over and over, tens of thousands of times a day, automating a workflow so thoroughly that the API logs became their own confession.

Internal documents reviewed by journalists don't reveal the employee's name or exact role. What they reveal is a pattern: a single person at a 200,000-employee company quietly rewired their entire workday around an AI model, running calls at a volume that would have required a small team of humans to match. Disney isn't publicizing this. The docs leaked it. Which means the same thing is almost certainly happening at dozens of other large companies — insurance firms, media conglomerates, law offices — quietly, without press releases.

The real jaw-dropper isn't the number. It's that no one needed approval. The employee didn't go through IT procurement or a committee. They just found a tool that worked and ran it into the ground. That's how automation actually spreads — not from the top down, but from one person's spreadsheet out.

Gobble's Take: Your job's automation isn't coming in a company-wide announcement — it's already happening in the cubicle next to yours, and nobody's sending a calendar invite.

Source: Reddit

Kevin O'Leary Just Got Approved to Build a Data Center That Eats More Power Than All of Utah

The state of Utah uses roughly 40 billion kilowatt-hours of electricity per year. Kevin O'Leary's newly approved AI data center campus will generate — and consume — more than twice that. All by itself. On one patch of land.

The 9-gigawatt campus, now cleared for construction, is one of the largest single data center projects ever approved in the United States. To put 9 gigawatts in context: that's roughly nine nuclear reactors worth of continuous power demand, dedicated entirely to running AI workloads. The project is part of a broader infrastructure arms race as hyperscalers and AI companies scramble to secure the compute capacity to train and serve next-generation models. Land, power, and cooling are now the scarcest commodities in tech — not talent, not capital.

Utah's grid will need to expand significantly to support the campus, raising questions about what happens to residential and commercial power rates in the region. The approval process apparently moved fast enough that public debate barely had time to catch up.

Gobble's Take: The AI boom's dirty secret isn't bias or hallucinations — it's that every chatbot answer you get costs the power grid about what it takes to charge your phone, multiplied by billions.

Source: Reddit

Wall Street Is Spooked: AI Fears Are Quietly Reshaping How Investors Bet on Growth

For a decade, the trade was simple: buy tech, hold, grow rich. Goldman Sachs is now warning that AI uncertainty is breaking that playbook.

According to Goldman, US stock investors are rethinking long-term growth bets specifically because of AI disruption risk — not the optimistic kind, but the kind where entire business models get quietly hollowed out before anyone notices. The concern isn't that AI fails. It's that AI succeeds in the wrong places: automating the revenue streams of companies that looked bulletproof two years ago. Software firms with fat margins built on human-labor bottlenecks are suddenly exposed. The analysts who once screened for "durable competitive moats" are now asking a harder question — does this company's moat survive a world where the task it charges for can be replicated for fractions of a cent per API call?

Goldman's note doesn't name names, but the implication is clear: the next wave of market volatility won't be triggered by interest rates or geopolitics. It'll be triggered by an earnings call where a CFO quietly mentions that AI replaced a revenue line nobody thought was replaceable.

Gobble's Take: The market already priced in AI as a winner — it forgot to price in who AI is winning against.

Source: Reuters

Logitech's New Dial Puts Microsoft Office Controls in Your Left Hand

Logitech's MX Creative Console — a physical dial-and-button pad that sits to the left of your keyboard — now works natively with Microsoft Office apps including Word, Excel, and PowerPoint.

The integration means you can map the dial to zoom, scroll, font size, or undo history, and assign the buttons to formatting shortcuts you'd normally chase through menus. It's a small hardware bet on a contrarian idea: that not everything about productivity should live on a screen. The MX Creative Console was originally aimed at video editors and designers who already use tools like Adobe Premiere and Lightroom. The Office expansion pushes it squarely into the knowledge-worker market — the hundreds of millions of people who live in spreadsheets and slide decks.

At $99, it's not an impulse buy, but for anyone who's ever lost three minutes hunting through Excel's ribbon for a formatting option, the pitch is easy to understand.

Gobble's Take: We've spent 30 years making software more powerful and keyboards exactly the same — a dial that finally talks to Excel feels less like a gadget and more like an apology.

Source: The Verge

Quick Hits

• Open-source AI is closing the gap — and the frontier keeps moving: A growing debate in AI circles centers on whether open models will eventually catch the closed frontier, or whether proprietary labs will always stay a step ahead — with real consequences for who controls the technology. Reddit
• Wall Street's three AI stock picks for 2025: Yahoo Finance rounded up analyst recommendations on the AI names most likely to outperform — useful context if Goldman's caution above has you reconsidering your portfolio. Yahoo Finance

In Case You Missed It

Yesterday's top stories:

• Optical Tornadoes Send Quantum Data 100x Faster — and Your Network's Worst Enemy Is Already Obsolete
• Musk's OpenAI Lawsuit Is Now AI's Biggest Ethics Test — and Every Lab Has a Stake in the Verdict
• The Nasdaq Is 15% Off Its Low and Racing Back Toward a Record — AI's "Dip" Window May Already Be Closing