The Model Y Problem: Why AI Won't Take Your Job as Fast as the Headlines Claim

The best product on the planet still can't get into most garages, and that same adoption friction is what gives the economy room to absorb AI - unless a slow bleed pushes unemployment to 7% and the ballot box turns.

Here's the thing I keep coming back to. Everyone is now terrified that AI is about to erase their job overnight, and I think the overnight part is almost certainly wrong. Not because the technology isn't good enough. Because being the best has never been enough to get adopted fast. The Tesla Model Y is, by almost any objective measure, the best car in the world, and most people with a garage still don't own one. That gap between "objectively superior" and "actually adopted" is the most underrated force in this entire conversation, and it's the reason I think the labor transition will be slower, messier, and more survivable than the scariest projections suggest. The real danger isn't speed. It's a slow bleed that lasts just long enough to break the politics.

Key Takeaways

• A year ago almost nobody was seriously debating AI's effect on jobs. Twelve to eighteen months later it's the loudest anxiety in the room, and I think that panic is quietly useful - it forces the conversation we needed.
• Self-driving is the clearest near-term win. Get autonomy to something like sub-60-cents-per-mile and the benefit becomes so obvious that even a scary headline can't override what people feel in their own lives.
• The Model Y is arguably the best car ever built - safety, range, software, comfort, reliability - and it still holds nowhere near a monopoly. Being the best doesn't hand you the market.
• AI is a product and a service, not magic. It faces the same thousand small frictions every great product faces, which stretches the adoption timeline far longer than the hype implies.
• Slower adoption is a feature, not a bug: it lets the economy absorb the shock. The nightmare scenario is a prolonged middle stretch where layoffs outrun new job creation and unemployment ticks from 5% to 6% to 7%.
• Sustained high unemployment is where politics breaks - that's when voters start reaching for socialist or worse policies at the ballot box.
• A U.S. sovereign wealth fund buying stakes in NVIDIA, Tesla, SpaceX, OpenAI, Anthropic and Meta at today's valuations could hand taxpayers a 10x-100x asset base in twenty years, modeled on Alaska's oil-funded dividend.
• Canada already runs a version of this: a national pension pool near $80 billion, funded by a 4% employee and 4% employer match, independently invested at roughly 13% annual returns for two decades.

---

The Panic Arrived Fast, and I Think That's Good

For a stretch there was this strange silence. AI was advancing hard and almost nobody in the mainstream was reckoning with what it meant for work. Then, in the space of a year to eighteen months, the mood flipped straight to dread. Robots are coming for you. You're obsolete.

My instinct says that fear, as ugly as it is, is actually healthy. It drops the discussion into the public square early, while there's still time to steer. You can't navigate a transition nobody's willing to talk about. The panic bought us the conversation.

The Media Landmine

There's a trap on the way through, and it's obvious once you name it. Fear sells. Media companies are structurally rewarded for running the most terrifying framing they can get away with, because terror moves clicks. So the loudest signal you'll hear about AI and jobs will be systematically tilted toward doom, not because anyone's lying exactly, but because that's what the incentive pays for.

The job, then, is to not mistake the volume for the truth. Once you route around the scariest-narrative machine, the actual picture gets a lot more workable.

Wins People Can Feel

I think there's only one way to get to the other side of this cleanly. The technology has to deliver big, visible wins fast - wins ordinary people feel in their own lives, not wins buried in a research paper.

Self-driving is the obvious candidate. If autonomy scales quickly and cheaply, something on the order of sub-60-cents-per-mile, the benefit stops being abstract. Your commute gets cheaper, safer, and you get your time back. And yes, it takes driving jobs. Absolutely it does, and we have to take care of the people it displaces. But that becomes a far easier conversation once the upside is sitting right in front of everyone. When someone reads a headline screaming that the robot is coming to kill them or steal their livelihood, and their actual daily experience says my life got better because of this, the headline loses. That cognitive dissonance is exactly what people need to climb across the gap.

The Model Y Tells You Everything

Now the part I find genuinely clarifying. Take the Model Y, a car we follow closely. Outside of China's fierce competition, I'd argue it's the best car on the planet and it isn't close. The safety numbers are absurd. It drives itself. It's quiet, quick, comfortable, reliable, well-packaged, great software, plenty of space. Go down the list and it keeps winning.

And it does not have a monopoly. Not remotely. The number of Model Ys on the road is still small, and Tesla isn't even production constrained - they could sell more tomorrow. So here's the uncomfortable lesson for the whole AI story: being objectively the best, and being obviously the best, does not get you universal adoption. Not even close.

AI, by default, will offer incredible value - great output for very low relative cost. That's precisely what the Model Y already is in its category. If the best car in the world can't get into most garages, why would we assume the best software in the world instantly gets into every workflow?

Why Slower Is Safer

For years I banged my head against the wall on this. Everybody with a garage should own a Tesla. It's objectively better, it saves money, it's a strictly superior product. And they don't - for a thousand reasons that are outside my control and outside Tesla's control, and plenty of those reasons are perfectly rational for the individual.

AI runs into the same wall. It's ultimately a product or a service. It will save people time. It will be enormously valuable. But will people treat it as adopt-this-or-you-die urgent? I don't think so, because people just don't behave that way. And here's the payoff: that friction, frustrating as it is, is what saves us. If AI diffuses slowly, the economy gets time to rebalance. Jobs disappear, new ones form, and the slower the clock runs, the better the absorption. The friction I used to curse is the shock absorber.

The 7% Trap

But slow isn't automatically safe, and this is the part I actually worry about. Picture the middle stretch. AI is laying people off faster than it's creating new roles, and the displaced can't find work quickly enough. Unemployment ticks from 5% to 6% to 7%.

If that holds for a while, it stops being an economic problem and becomes a political one. Prolonged joblessness is how you manufacture instability. That's when people walk into the voting booth angry and start electing whoever promises to burn it down - socialist policies, worse policies, the whole menu of desperation. The technology could be working exactly as designed and still hand us a political crisis through the sheer duration of the gap. That middle section is the real risk, not the robots.

A Sovereign Wealth Fund for the AI Boom

So what do you actually do about the money question, since the honest anxiety underneath all of this is simple: where does my income come from if I'm laid off? Elon likes to say money won't be needed in the future. Maybe. That's a very long time out, and I agree with the skeptics that it's a long way off. People need income now.

One answer I keep circling is a U.S. sovereign wealth fund. The country takes equity positions in the companies driving this boom - NVIDIA, Tesla, SpaceX, OpenAI, Anthropic, Meta - at today's valuations. Twenty years out, if those stakes 10x or 100x, the government is sitting on an enormous asset base. It can sell and redistribute the gains, or rotate into dividend-paying assets and cut checks to citizens directly, the way Alaska's fund distributes oil proceeds. It turns the AI upside into cash flow for the taxpayer instead of a windfall that accrues only to the people who already own the shares.

I won't pretend it's clean. The obvious objection is that government starts picking winners, and yes, it will get corrupted - lobbyists, handshake deals, someone squeezing someone else for the allocation. But we already do exactly this in defense. Why Lockheed, why Northrop, why SpaceX for this contract and not that one? We live in an imperfect world, so the real question is which imperfect mechanism beats the alternatives. On that test, a sovereign wealth fund is one of the better ideas I've seen.

Canada Already Built a Version of This

Here's the model that makes me think it's more than theory. Canada doesn't have a sovereign wealth fund in the American sense, but its national pension isn't a government piggy bank either. It's a large pool, somewhere around $80 billion, independently managed and actually invested on behalf of the population. Workers pay in roughly 4%, employers match another 4% up to a limit, and the pool has reportedly compounded at something like 13% a year over the past couple of decades.

That structure does what a pay-as-you-go system can't. It's liquid, it can absorb a population boom, and managed properly it doesn't go bankrupt. The same logic is exactly why the idea of quote-unquote privatizing Social Security into an invested, sovereign-wealth-style vehicle keeps gaining traction. We are very obviously in a capital-deployment boom right now. Enormous sums are flowing into the very companies these programs could be holding. For the pools meant to protect people through a transition like this one to sit on the sidelines while that value compounds strikes me as close to negligent. There's a real opportunity in front of us. The only question left is whether we're organized enough to grab it.