The Surveillance Bet: Big Tech's Desperate Pivot to AI Dominance

---

Meta’s going to watch you type.

Not your customers. You. The person sitting in the office wearing the company badge. Every keystroke, every mouse movement, fed into the machine learning pipeline—all in service of training better AI models. This isn’t a privacy policy footnote. This is a company basically saying: we’re so desperate to win the AI race that we’re willing to turn our own workforce into training data.

That headline landed last week like it barely registered. It should’ve landed like a bomb.

Photo by David Yu / Pexels

The Desperation Is Showing

Here’s the thing about the current AI moment: everyone’s terrified they’re already losing. Meta’s keystroke tracking, Amazon’s $25 billion bet on Anthropic, SpaceX’s $60 billion acquisition of Cursor (which Elon’s clearly positioning as an AI play ahead of going public)—these aren’t confidence moves. These are panic moves. The game changed faster than anyone expected, and now every major tech player is throwing money and ethics at the wall to see what sticks.

Amazon committing to spend $100 billion on its own infrastructure alongside the $25 billion investment in Anthropic? That’s not diversification. That’s hedging a bet you’re not sure you’ll win. The company’s essentially saying: we don’t trust ourselves to build this alone, but we also can’t afford not to try.

Meta’s surveillance angle is actually the most honest expression of this fear. They’re not pretending they’re innovating their way to AI supremacy. They’re admitting: we need data, we need it fast, and the easiest source is internal. Your fingers on the keyboard are just another training set.

The cynical read is obvious. The smarter read is sadder: Meta’s probably right that this is their best play. The companies winning at AI right now aren’t the ones with the biggest server farms—they’re the ones with the most usable training data and the willingness to exploit every possible source of it.

Photo by nappy / Pexels

Apple’s Gamble on “Product Guy” Energy

Meanwhile, Apple’s replacing Tim Cook with John Ternus, a hardware engineer who’s never run a company. Cook was exceptional at one thing: making money. The Steve Jobs era was about innovation; the Cook era was about optimization and extraction. Gross margins, services revenue, the entire ecosystem lock-in strategy—that was Cook’s masterpiece.

Ternus is walking into a company that’s phenomenally profitable but creatively stalled. The last genuinely interesting Apple product was the Apple Watch, and that’s now almost a decade old. iPhone? It’s a telephone. iPad? A bigger telephone. The Vision Pro exists, and we’re all pretending it’s the future while it sits on shelves gathering dust.

My read: Ternus is being asked to do something much harder than making money. He’s being asked to make people care again. Whether a hardware guy can do that when the entire industry’s being reshuffled by AI is the question. Apple’s been weirdly quiet on AI strategy compared to everyone else. Is that caution or complacency? Ternus has to answer that in his first year, and I’m genuinely uncertain which way he’ll go.

The Fringes Are Getting Weird

Peter Molyneux—yes, that guy, the creator of Fable—says his next game, Masters of Albion, will be his last. He’s old. Fair enough. But the fact that a legend’s signing off right as AI’s reshaping game design? That feels like a punctuation mark. Molyneux was always about systemic depth, player choice, consequences. AI could theoretically enable that at scale. Instead, he’s walking away.

Then there’s the UK’s National Cyber Security Centre basically endorsing “frontier AI hacking tools” like Mythos as potentially positive if kept out of the wrong hands. This is a government agency saying: yes, we’re okay with AI-powered hacking tools existing in the world, we just trust ourselves to be the right hands.

That’s either visionary or delusional. Possibly both.

And Telegram’s under investigation for child sexual abuse material while denying everything, which is a separate crisis but tells you that growth + encrypted communication + minimal moderation = problems that no amount of AI surveillance can fix.

Here’s What Actually Matters

The real story isn’t any single headline. It’s the pattern: every major tech company is making a bet that AI is worth compromising on everything else for.

Meta’s willing to watch workers. Amazon’s willing to pour $125 billion into a bet that might not pay off. Apple’s willing to hand the wheel to someone untested in crisis leadership. SpaceX is positioning itself as an AI company now, not a rocket company.

This is how industries shift. It’s not elegant. It’s messy and desperate and full of companies doing things they’d normally never consider, all because they’re terrified of being the one left behind. In 2012, every company thought they needed an Instagram clone. In 2015, it was machine learning talent. In 2018, it was blockchain. By 2025, it’s clear: the real battle is for training data and inference capacity.

The surveillance piece—Meta’s keystroke tracking—is just the logical endpoint of that desperation. If your training data is your competitive advantage, and you’ve already scraped the internet, mined social media, and bought up datasets, the next frontier is the stuff you generate internally. Your employees’ work patterns become product.

I think this backfires within three years. Not for Meta specifically, but as a broader trend. There’s a talent cost to treating workers as training data. There’s a creativity cost to being watched. And there’s a regulatory cost that’s coming—Europe’s already sharpening knives, and the U.S. won’t be far behind.

But that’s a future problem. Right now, in Q1 2025, the bet is still working. Money’s still flowing. Stock prices are still climbing.

What I’m Watching

• Meta’s talent attrition rate through 2025. If keystroke tracking is the policy, watch whether senior engineers start leaving. That’s the real tell of whether this bet works internally.

• Amazon’s Anthropic stake breakdown by Q2 2026. Did the $25 billion investment actually improve AWS’s AI offerings relative to competitors? Tangible metrics—latency, cost, adoption—matter here.

• John Ternus’s first major product announcement. He’s got maybe 18 months before the market expects something genuinely new from Apple. If it’s an incremental refresh, the “product guy” narrative collapses fast.

• Whether the Telegram investigation moves to enforcement. Ofcom investigating is theater unless it leads to actual penalties. Watch for a ruling by mid-2026.