From Critic to Believer

There's something satisfying about watching someone publicly change their mind with money on the line. Serial AI entrepreneur Pieter Levels, who operates under the handle Levelsio, announced Thursday on X that he'd purchased over $1 million in Alphabet Inc. (GOOG) shares. The kicker? He's been, by his own admission, "the biggest Google hater for years."

So what changed? According to Levels, pretty much everything. He argued that Google had been "completely mismanaged" for years and "destroyed by politics and lack of any leadership." But then came a turnaround that he credits to co-founder Sergey Brin's return as what Levels calls the "de facto CEO."

Now, Levels says, Google isn't just putting up impressive numbers on AI leaderboards (which he notes "can, and are easily rigged"). The company is "dominating not just in the AI benchmarks and leaderboards but in real usage." That's the distinction that matters when you're writing seven-figure checks.

The Hardware Advantage

Levels laid out his thesis with the kind of specificity that suggests he's done his homework. Google's proprietary tensor processing units, the massive data moat built through YouTube, Google Images, and web search, plus what he sees as renewed engineering leadership under Brin. These are the ingredients he thinks will carry Google to the front of the AI race.

"It's really just Google and Elon Musk and the Chinese in the end who will probably win," Levels wrote, noting that he now only relies on models from Google, Musk's xAI, and Chinese companies like ByteDance, Kling, and Minimax.

Levels, who founded PhotoAI, InteriorAI, and RemoteOK, also mentioned he'd trimmed his NVIDIA Corp. (NVDA) holdings months earlier. His reasoning? Competition in AI chips was "inevitable," and Google's TPUs now present a credible alternative.

Musk's Cryptic Response

Enter Tesla Inc. (TSLA) CEO Elon Musk, who responded to Levels' lengthy post with exactly six words: "Hardware is the end game."

It's classic Musk brevity, but there's substance behind it. Tesla has been scaling its in-house AI chip program aggressively. Last month, Musk made a characteristically bold claim: "We expect to build chips at higher volumes ultimately than all other AI chips combined. Read that sentence again, as I'm not kidding."

Whether that's realistic is debatable, but it signals where Musk thinks the real competition will happen. Not just in algorithms or training data, but in the physical infrastructure that makes AI possible at scale.

Dan Ives, a senior analyst at Wedbush, has called Tesla one of the market's leading "physical AI plays," pointing to the company's efforts to embed AI into tangible systems like autonomous vehicles, robotaxis, and the Optimus humanoid robots.

Market Response

Investors seemed to like what they were hearing. Alphabet shares climbed 1.91% on Thursday, closing at $303.75, and ticked up another 0.47% in after-hours trading. Tesla gained 3.45% to close at $483.37, with an additional 0.43% overnight bump.

The broader narrative here is about vertical integration. As AI becomes more central to tech giants' business models, controlling the entire stack—from chips to data centers to end-user applications—looks increasingly like a competitive necessity rather than a luxury. Google has its TPUs. Tesla is building its own chips. Even companies like Amazon and Microsoft are developing custom silicon.

Levels' investment, combined with Musk's pointed comment, underscores a shift in how serious players are thinking about AI. The software is important, sure. But the hardware underneath might be where the real moats get built. And when someone who's been vocally skeptical for years suddenly puts a million dollars behind that thesis, it's worth paying attention.