The Year Silicon Valley Stopped Believing in Its Own Hype

---

Something shifted. Not gradually. Not in some abstract way you’d need a think tank to measure. It’s visible in the lawsuits, the settlements, the sudden interest in oversight from people who’d previously treated regulation like a contagious disease.

The tech industry is being forced to reckon with the idea that maybe, just maybe, you can’t innovate faster than you can be held accountable.

When the Lawsuit Becomes the News

Start here: Elon Musk just paid the SEC $1.5 million to settle charges that he hid his Twitter stock purchases. That’s not a scandal in the traditional sense. It’s a guy paying a fine. But consider the context.

The SEC has been pulling back on lawsuits against major companies. It’s the agency’s actual policy right now—less enforcement, more negotiation. So when Musk still gets caught, when it still costs him money, you’re watching what happens to the guy who’s supposed to be untouchable. The richest man on Earth is being nickeled-and-dimed by bureaucrats.

Meanwhile, in federal court, Musk’s lawyers are grilling OpenAI’s president Greg Brockman about why he’s worth $30 billion. Not “is this a fair valuation”—but implying, directly, that Brockman was driven by greed rather than the mission of building safe AI. Brockman’s response: he felt threatened. “I thought he was going to hit me.”

This is the co-founder trial between Musk and OpenAI. It’s supposed to be about whether OpenAI betrayed its nonprofit mission when it pivoted to a capped-profit model. But what it’s actually revealing is that these guys don’t like each other. That the whole thing was fractured from the start. That billions of dollars and the future of artificial intelligence depend on people who have, apparently, had moments where physical violence seemed possible.

Photo by Robert So / Pexels

The Government Woke Up

Here’s my read on what just happened: the Trump administration, which everyone assumed would be pure hands-off on AI, is now considering pre-release vetting of AI models. Not after they’re causing problems. Before.

This is a 180-degree turn from “move fast and break things.” This is “move slow and have someone check your work first.”

The Biden-era pacts between companies and the Commerce Department already created a framework. Google, Microsoft, and xAI are now participating in safety testing agreements. It’s voluntary, technically, but voluntary is doing a lot of heavy lifting here. The moment a major AI lab opts out, you’ve got a regulatory opening the size of a stadium. So they’re all in.

I think what happened is simple: Ukraine showed us what unmanned warfare looks like when Zelensky announces that territory was captured using just robots and drones. Not robots and soldiers. Not drones supporting humans. Robots. Territory control by machines.

That’s not a philosophical concern anymore. That’s a Tuesday. The future of unmanned warfare isn’t a debate at a conference—it’s happening in real time, and governments are watching. When governments watch, they regulate.

The Money Keeps Coming (With Strings)

Anthropic and Wall Street giants are launching a new AI firm. Blackstone. Goldman Sachs. These aren’t venture capitalists anymore—these are the institutions that know how to navigate regulation. They’re betting that the future of AI isn’t some scrappy startup in San Francisco. It’s an enterprise play, embedded in financial systems, designed from the start to survive congressional testimony.

That’s a structural shift. Capital doesn’t move that way without a reason.

The Consumer Revolt Isn’t Coming. It’s Here.

Apple’s paying up to $95 to some US iPhone buyers because it lied about Apple Intelligence in its advertising. Not a massive settlement, but significant for one reason: it established that you can’t just claim your AI does things it doesn’t do. There’s now legal precedent.

And then there’s Pornhub—yes, Pornhub—becoming accessible again in the UK for iPhone users who complete Apple’s age verification. This is a weird detail that matters. Apple’s privacy-first, device-based age checking isn’t being used to block access. It’s being used to allow access. The infrastructure cuts both ways.

Photo by nappy / Pexels

The One Factory That Gets It

Here’s what should terrify every other manufacturer: Bristol Myers Squibb’s plant in the US is the only one recognized by the World Economic Forum this year for AI innovation in manufacturing.

One factory.

In the entire country.

American factories are lagging in AI adoption. This isn’t about being first—it’s about being last. It’s about competitive advantage narrowing to a thread while everyone else figures out why their supply chains haven’t improved in five years.

What This All Means Together

None of these stories are accidents. They’re interconnected threads:

The warfare angle forces governments to care. Once governments care, they regulate. Once regulation exists, capital becomes risk-averse. Once capital is risk-averse, the wild-west startup phase ends. Once that ends, you’re left with incumbents (Goldman, Blackstone, major tech platforms) building defensible positions.

The Musk-OpenAI trial is the personal cost of that transition. The settlement is the institutional cost. The safety tests are the operational cost. The advertising lawsuit is the consumer protection cost.

My prediction: by Q3 2025, we’ll see the first major AI company pivot away from public release toward enterprise-only deployment. They’ll frame it as “safety.” It’s actually about margin protection and regulatory risk. The shift from “AI for everyone” to “AI for approved partners only” is coming. Maybe not at OpenAI—they’re too public-facing. But somewhere in the top five labs, you’re going to see that move.

Here’s my genuine uncertainty, though: I don’t know if the pre-release vetting actually works. Do safety tests catch dangerous models? Or do they just slow deployment and let testing become theater? We won’t know until someone pushes back, and pushback requires a lab confident enough to resist. I don’t see that lab yet.

One more thing I’m watching: that Pornhub detail. If Apple’s age verification works for adult content, someone’s going to demand it work for other things. Gambling. Financial products. Alcohol. That infrastructure could become the template for how we gate digital access generally. Could go either way—privacy nightmare or reasonable guardrail. But that system is going to be tested hard.

What I’m Watching

• The first enterprise-only AI lab announcement. Watch for a major AI company announcing they’re restricting access, claiming safety. My bet: Q2-Q3 2025. This marks the actual turning point from “tech for everyone” to “tech for approved businesses.”

• How many factories adopt AI by end of 2025. That Bristol Myers Squibb plant is a lonely datapoint. If US manufacturing doesn’t move significantly, you’re watching competitive decline in real time. Watch WEF’s next innovation recognition list.

• Whether xAI actually participates in safety testing. Musk’s company signed on to Commerce Department agreements. Will they submit to pre-release vetting? Or will they break formation first? That’s the real test of whether this regulatory framework has teeth.

• The settlement amount in the next Apple AI lawsuit. $95-per-person is a rounding error. But if the next one is significantly higher, you’ll know courts are taking AI advertising seriously. That changes the incentive structure overnight.