The Great Silicon Valley Handoff: Who Actually Controls Tech Now?

---

Tim Cook made Apple the most profitable company on Earth. Now he’s leaving.

That sentence should land harder than it does. For 14 years, Cook turned Steve Jobs’ innovation machine into a cash printing operation—a shift so complete that calling it a “transition” undersells the magnitude. The Tim Cook era wasn’t about inventing the next revolutionary product. It was about making the iPhone, iPad, and Mac so financially optimized that Apple became untouchable. Record margins. Insane returns. A business so locked-in that investors stopped caring whether Apple would ever innovate again.

Enter John Ternus, hardware engineer, company lifer, and the guy who actually built the stuff Cook monetized.

This matters because it signals something bigger than one CEO departure. The entire tech industry is in the middle of a generational handoff, and we’re about to see whether the people running things tomorrow actually understand what they built—or if they’re just inheriting the playbook.

Photo by Robert So / Pexels

The Cook Legacy Is Dead. Now What?

Let’s be clear about what Cook accomplished: he ran a company valued at $3 trillion while maintaining profit margins that would make most Fortune 500 CEOs weep. He did this by perfecting supply chain efficiency, squeezing every penny from services, and making you buy the same product in slightly different sizes every year. It’s impressive in the way that a perfectly executed factory system is impressive.

But here’s the thing about perfect systems: they don’t adapt well when the game changes.

Apple needs new ideas. Not incremental updates. Not another color option. The company Ternus is inheriting is extraordinarily profitable and extraordinarily stuck. Everyone’s wondering if a hardware guy who’s spent his career inside Apple can actually think beyond the box his predecessor built. Ternus ran hardware engineering. He knows how to execute a vision. Whether he can create one is still an open question.

My read: Ternus gets maybe 18 months of grace period before Wall Street decides if he’s Cook 2.0 or actually different. The first move matters disproportionately.

When Companies Start Watching Employees Like Prisoners

Now flip to Meta, where Mark Zuckerberg just announced the company will track workers’ clicks and keystrokes to train AI models.

This is dystopian and practical in equal measure. Meta’s betting that the data generated by how thousands of engineers actually work—the patterns, the decisions, the mistakes—can be converted into training data for AI. It’s Orwell meets optimization. And it’s probably going to work.

Here’s what’s genuinely unnerving: this isn’t surveillance disguised as productivity management. It’s explicit. Meta told employees this is happening. They’re not hiding it. That’s almost worse because it suggests the company believes workers have no practical alternative. You want to work at Meta? We watch everything. Deal with it.

I think we’re watching a template get tested. If Meta’s AI training improves measurably from this, every other tech company will adopt it. Amazon, Google, Microsoft—they’re all watching. Within two years, this could become industry standard. And nobody will unionize over it because the workers are already salaried tech employees with stock options, and they’ve been trained to accept this as the cost of being elite.

The prisoners who don’t have internet access but are finding ways to use chatbots anyway? They’re actually freer than Meta employees right now.

Photo by nappy / Pexels

The AI Arms Race Just Got Weird

SpaceX buying Cursor for $60 billion is the kind of move that sounds insane until you realize Elon Musk is preparing to take his rocket company public and needs a killer AI story. A $60 billion acquisition for a coding tool isn’t about the tool. It’s about the narrative: SpaceX is positioning itself as an AI-first aerospace company, not a spacecraft company that uses AI.

That’s a rebranding play disguised as a technology bet.

Meanwhile, Amazon’s dropping up to $25 billion on Anthropic, and Anthropic’s committing $100 billion to use Amazon’s infrastructure. This is less a transaction and more a formal alliance. Amazon gets equity and influence in a top-tier AI lab. Anthropic gets guaranteed capital and computational resources. Both companies get to tell their investors they’re really serious about this AI thing.

The subtext? Google, Microsoft, and OpenAI are moving so fast that everyone else is in panic-acquisition mode. You can’t build an AI frontier lab fast enough anymore, so you buy influence instead.

When Hacking Tools Become Legitimate

The UK’s top cyber official just said AI hacking tools like Mythos can be a “net positive” if they’re kept out of the wrong hands.

That’s… honest. And terrifying. We’re one escaped model away from seeing this tested in real-time. The idea that the same AI that can find vulnerabilities in software can find vulnerabilities in, say, nuclear facilities or financial systems—and that the difference between “net positive” and “catastrophic” is just gatekeeping? That’s not actually a reassuring take.

But here’s what I actually think: the official’s right, and we’re naive if we expect gatekeeping to hold. AI hacking tools will proliferate. The question isn’t whether the wrong hands will get them. The question is how fast security can adapt once they do.

The Outliers and the Uncomfortable Ones

Peter Molyneux—legendary game designer, creator of Fable—just announced Masters of Albion will be his last game. After a career spent imagining worlds, he’s done. That’s either a sign that traditional game design is exhausted, or that the economics of making games for human audiences has become unworkable, or both.

Ofcom’s investigating Telegram over child sexual abuse material. This one’s straightforward: platforms are going to get scrutinized harder for what happens inside them, AI or not.

These aren’t the flashy stories, but they’re the ones that’ll matter in 18 months.

Photo by Markus Spiske / Pexels

What I’m Watching

• John Ternus’s first AI/innovation announcement — Within his first year as CEO, does Ternus commit Apple to a major AI initiative that’s separate from Cook’s playbook? The substance matters more than the spin. A real strategic pivot (not a feature add) signals whether this is actually generational change.

• Meta’s AI training improvement metrics — Six quarters from now, does Meta report measurably better AI model performance correlated to employee keystroke data? If yes, surveillance-as-training becomes industry standard by 2026. If no, it becomes a PR cautionary tale.

• SpaceX’s public filing narrative — When SpaceX goes public (likely 2025-2026), does the company lead with AI capabilities or rocket technology? The emphasis tells us whether this is actually a bet on reusable rockets or a bet on Musk’s ability to convince investors that AI is the future.

• First significant AI hacking tool breach or misuse case — When (not if) an AI hacking tool gets deployed in the wild by malicious actors, how long does it take for defenders to adapt? Under 30 days means we’re fine. Over 90 days means we’re in trouble. This is the real test of whether gatekeeping works.