A Google engineer correctly predicted that a near-unknown singer named D4vd would be the most-searched person on Google in 2025 — at odds so long that Polymarket users essentially thought it was impossible. That alone should have raised eyebrows. Now federal prosecutors say it wasn't clairvoyance. It was a cheat code.
Michele Spagnuolo, a software engineer at Google, was arrested in New York this week and charged with commodities fraud, wire fraud, and money laundering. Prosecutors allege he used access to Google's internal marketing data to place winning bets on Polymarket — a crypto-based prediction market — under the username AlphaRacoon. His alleged haul: $1.2 million. He was released on a $2.25 million bond.
The mechanics of the alleged scheme are actually pretty clever, in a deeply illegal way. Google's "Year in Search" rankings aren't based on raw search volume — they measure which terms saw the sharpest spike in interest over the course of the year. That quirky methodology makes the results genuinely hard to predict from the outside. But if you work at Google and can access internal data showing which terms are trending before the public announcement? Suddenly those long-shot bets don't look so long-shot anymore.
Spagnuolo allegedly used that edge to bet correctly on D4vd's top ranking while also betting against bigger names like Pope Leo XIV and Kendrick Lamar appearing on the list. His winning streak was unusual enough that outlets like Forbes wrote about it last December. Turns out the story behind the story was even more interesting.
Prosecutors also allege that Spagnuolo didn't just win — he tried to cover his tracks, taking steps to obscure where the money came from. That detail matters because it suggests this wasn't some impulsive mistake. It was a deliberate, multi-step operation.
Polymarket responded by posting on X that its "market integrity infrastructure" had flagged Spagnuolo's activity and that the company is cooperating with law enforcement. The platform leaned hard into the idea that blockchain transparency makes it a tough place to hide. That's partially true — on-chain transactions are traceable — but the case also illustrates that the real vulnerability wasn't the blockchain. It was the human sitting on a trove of nonpublic information.
This arrest lands at an awkward moment for prediction markets broadly. Several states have been pushing for stricter oversight of platforms like Polymarket and Kalshi, specifically citing insider trading risks. The CFTC and the Trump administration have pushed back, arguing that federal regulators have exclusive authority here. This case hands ammunition to everyone who's been arguing that prediction markets are structurally exposed to exactly this kind of abuse.
For Google, the episode raises uncomfortable questions about internal data access controls. A spokesperson confirmed the employee accessed marketing material through a tool available to all employees — which, depending on how you read it, either suggests this was a systemic gap or that tighter access controls might have prevented everything. Either way, it's not a great look.
AI
Nvidia Pledges 150 Billion to Taiwan as US AI Hub Plans Falter
Jensen Huang didn't mince words. Standing at a ceremony in Taiwan this week, the Nvidia CEO declared that the island should remain the "epicenter" of the AI revolution — and backed it up with a staggering commitment: $150 billion a year in Taiwan investments. For context, that's more than most countries spend on their entire defense budgets.
The announcement is significant not just for its scale but for its timing. Huang made it while the Trump administration is actively pressuring American tech companies to build on US soil and position the country as the world's AI manufacturing powerhouse. Nvidia hasn't publicly explained how a $150 billion annual commitment to Taiwan squares with that agenda. The company didn't respond to questions about the apparent tension.
To understand why Taiwan is so hard to quit, you have to understand what actually happens there. It's not just chip fabrication — Taiwan handles the advanced packaging process that binds chips into the high-performance systems AI workloads demand. That step is extraordinarily difficult to replicate elsewhere, and it's the reason Nvidia's supply chain still runs through Taipei even as the company breaks ground on US facilities.
Huang acknowledged that Nvidia's Taiwan spending has exploded over the past several years — from roughly $10 to $15 billion annually just four or five years ago, up to $100 billion now, heading to $150 billion. The new Taiwan headquarters, expected to be operational by 2030, is meant to lock in that relationship for decades and keep Nvidia at the center of what Huang is calling a new era of "agentic AI."
None of this means Nvidia is abandoning the United States. Last year, the company began producing AI chips on American soil for the first time — a move widely read as an olive branch to the Trump White House. Huang projected at the time that Nvidia could produce up to half a trillion dollars of AI infrastructure in the US over four years. Those plans are presumably still on the table.
But there's a hierarchy becoming visible here. The US investments are real, but they exist within the constraints of what's actually manufacturable domestically. Taiwan is where the full capability lives right now, and where Nvidia is choosing to plant its flag most firmly for the future.
The broader AI infrastructure buildout only amplifies the stakes. Tech giants are collectively on track to spend around $750 billion on AI infrastructure this year alone, with chips representing a massive slice of that. Whoever controls the most advanced chip packaging and production capacity holds enormous leverage over that entire ecosystem — and right now, that's Taiwan.
For the Trump administration, Huang's announcement is a quiet but pointed reality check. Reshoring semiconductor manufacturing is a worthy long-term goal. But the infrastructure, expertise, and supply chain depth concentrated in Taiwan took decades to build. Nvidia is betting it'll take at least a few more to replicate anywhere else.
Source: Ars Technica
The announcement is significant not just for its scale but for its timing. Huang made it while the Trump administration is actively pressuring American tech companies to build on US soil and position the country as the world's AI manufacturing powerhouse. Nvidia hasn't publicly explained how a $150 billion annual commitment to Taiwan squares with that agenda. The company didn't respond to questions about the apparent tension.
To understand why Taiwan is so hard to quit, you have to understand what actually happens there. It's not just chip fabrication — Taiwan handles the advanced packaging process that binds chips into the high-performance systems AI workloads demand. That step is extraordinarily difficult to replicate elsewhere, and it's the reason Nvidia's supply chain still runs through Taipei even as the company breaks ground on US facilities.
Huang acknowledged that Nvidia's Taiwan spending has exploded over the past several years — from roughly $10 to $15 billion annually just four or five years ago, up to $100 billion now, heading to $150 billion. The new Taiwan headquarters, expected to be operational by 2030, is meant to lock in that relationship for decades and keep Nvidia at the center of what Huang is calling a new era of "agentic AI."
None of this means Nvidia is abandoning the United States. Last year, the company began producing AI chips on American soil for the first time — a move widely read as an olive branch to the Trump White House. Huang projected at the time that Nvidia could produce up to half a trillion dollars of AI infrastructure in the US over four years. Those plans are presumably still on the table.
But there's a hierarchy becoming visible here. The US investments are real, but they exist within the constraints of what's actually manufacturable domestically. Taiwan is where the full capability lives right now, and where Nvidia is choosing to plant its flag most firmly for the future.
The broader AI infrastructure buildout only amplifies the stakes. Tech giants are collectively on track to spend around $750 billion on AI infrastructure this year alone, with chips representing a massive slice of that. Whoever controls the most advanced chip packaging and production capacity holds enormous leverage over that entire ecosystem — and right now, that's Taiwan.
For the Trump administration, Huang's announcement is a quiet but pointed reality check. Reshoring semiconductor manufacturing is a worthy long-term goal. But the infrastructure, expertise, and supply chain depth concentrated in Taiwan took decades to build. Nvidia is betting it'll take at least a few more to replicate anywhere else.
Enjoyed this?
Get stories like this delivered every Tuesday — free.