Tracking Wealth Through the AI Lens
Tracking Wealth Through the AI Lens
Michael Burry, the guy who called the 2008 housing crash, is now waving red flags about AI. Accusing hyperscalers like Google of accounting shenanigans? That’s a bold move, even for him. But here’s the thing: while everyone’s hyper-focused on the potential bubble, they're missing the real story. Google isn't just riding the AI wave; they're building the infrastructure that powers it, and that's a very different game.
Forget the day-to-day stock tickers and analyst squabbles for a minute. This isn't about whether GOOG will hit $300 next quarter. This is about the fundamental shift in how we compute, create, and connect. Think of it like this: Burry is worried about tulip mania, but Google is selling the fertile soil where all the flowers grow.
Burry’s accusation, in essence, boils down to this: hyperscalers are artificially inflating their earnings by understating depreciation on their AI infrastructure. He estimates a whopping $176 billion in understated depreciation between 2026 and 2028. He points a finger specifically at Alphabet, claiming they've doubled the "useful life" of their network/compute assets since 2020. Okay, fair point. But what's he missing?
He’s missing the why.
Google's not just extending the lifespan of old tech; they're innovating at a pace that redefines what "useful life" even means. The article mentions Alphabet's Tensor Processing Units (TPUs), and this is where it gets exciting. Anthropic, a major AI player, plans to buy one million of these TPUs. One million!
That's not just a revenue opportunity; that's validation of a whole new computing paradigm. It's like Ford not just building cars, but also inventing the assembly line and then selling that assembly line to other manufacturers. Suddenly, Google isn’t just a search engine or a cloud provider, but the Intel of the AI era, and that changes everything.
And here’s the kicker: the demand for these TPUs isn’t just some flash-in-the-pan trend. It's driven by the insatiable hunger of AI models for more processing power. As AI gets more sophisticated, the need for specialized hardware like TPUs will only skyrocket. Google's not just selling shovels in the gold rush; they're selling self-replicating shovels.
The numbers back it up. Google Cloud revenues jumped 34% YoY, hitting $15.2 billion, with a backlog soaring to $155 billion. These aren’t just incremental gains; they're exponential leaps.
Remember when everyone thought OpenAI would eat Google’s lunch? Google Search revenues rose 15% YoY. People are still searching, still clicking, still generating ad revenue. Google isn't just defending its turf; it's expanding its empire. Is GOOG Stock a Buy or Sell as Michael Burry Accuses Hyperscalers of ‘Fraud’?
I remember when I first read about the TPU project years ago. The sheer audacity of Google deciding to build its own custom chips was breathtaking. Most companies would have just bought off-the-shelf solutions. But Google saw something others didn't: that the future of AI demands specialized hardware, and they were willing to bet billions on it.
But where do we go from here? Will Google's TPU dominance stifle innovation from other players? What ethical considerations arise when a single company controls such a critical piece of the AI infrastructure? These questions need answers, and fast.
Burry's a smart guy, no doubt. But he's looking at the surface, at the froth of the market. He’s seeing the AI hype and assuming it's all smoke and mirrors. But Google’s building something real, something tangible, something that will outlast the hype cycle. They're not just playing the AI game; they're changing the rules. This isn't just about AI; it's about the future of computing, and Google is holding the keys.