Bill Gates published a 6,000 word essay last month on his personal site. It's called The turbulent AI era is here. Most people who read newsletters like this one saw the headlines, nodded, and moved on. The headlines were the boring part.

The interesting part is that Gates - who has more visibility into where AI is actually heading than almost anyone alive - just described our future. Not a vague one. A specific one, with dates, mechanisms, and consequences. Then a group of very smart people spent two weeks arguing with him in public, and their arguments made the picture sharper, not fuzzier.

If you have not had time to read either the essay or the responses, here is what you missed.

See below to continue…

The best marketing ideas come from marketers who live it. That’s what The Marketing Millennials delivers: real insights, fresh takes, and no fluff. Written by Daniel Murray, a marketer who knows what works, this newsletter cuts through the noise so you can stop guessing and start winning. Subscribe and level up your marketing game.

Sign Up Free

Continue…

Gates says smart robots will begin competing with humans on physical work - construction, hospitality, elder care - by the end of this decade. That is four years. Not fifteen. Four.

Most people I speak to still think of physical AI as a next-decade problem. Something to keep an eye on. Something the Chinese are doing but not really relevant to Europe yet. Gates has access to more robotics roadmaps than any of us. When he puts a specific date on the shift, that date deserves attention.

He is also blunt about why most people are getting this wrong. He says the viral videos of robots dancing badly have made everyone underestimate how fast dexterous manipulation is advancing, most of it in China, most of it not in English. If your only exposure to humanoid robotics is TikTok clips of Optimus falling over, you have been shown the wrong data.

The consequence is not that half the workforce disappears next year. The consequence is that the pricing calculus for a lot of physical work quietly inverts, then keeps inverting, and companies that spot the shift a year early will look, in retrospect, like they had inside information. They didn't. They just read the essay.

Employers have cited AI as the reason for 184,538 job cut announcements since 2023. In July of this year alone, AI was the leading stated reason for 10,970 layoffs. The 2026 running total is 112,713.

Gates points at a Stanford paper that goes further. After generative AI became widespread, employment fell significantly among young workers in the roles most vulnerable to replacement. Not among their older colleagues. Just the entry level.

This is the part that should worry anyone building a company that plans to hire graduates in the next three years. The workforce pipeline is being reshaped in real time. The people who would have been your first hires in 2028 are not entering the industries they would have entered. The junior analyst layer, the paralegal layer, the customer support layer, the coding bootcamp layer - all of it is thinning at the bottom while nobody quite says it out loud.

If you are a founder relying on the assumption that talent will be plentiful and affordable in 2027 because it always has been, Gates is telling you that assumption is broken. If you are a founder building a product whose value proposition is "replace your junior team with our AI," you are riding a very fast trend that Gates is now openly asking governments to slow down.

That last part is the second thing worth paying attention to.

Gates proposes taxing AI tokens and robots the way governments currently tax human labour. Hire a person, you pay payroll tax. Buy a robot or an AI system that does the same job, you write it off as a business expense. Gates argues this quietly subsidises replacing humans with machines, and he wants to reverse the incentive.

The response was immediate and largely negative. Jensen Huang, who runs Nvidia and therefore sells the hardware in question, said publicly he loves Gates but does not see what he sees. Huang's counter-argument is that productive companies hire rather than fire, and AI will be a net job creator at a scale never seen. Oren Etzioni of the Allen Institute pointed out a harder technical problem: a token tax can only be enforced at commercial API billing boundaries. Models running locally on owned hardware produce no invoice and cross no taxable line. The first-order effect of a token tax would therefore be to accelerate the migration to open-source local models and to push commercial demand toward Chinese alternatives that face no equivalent levy.

Both objections are correct. The tax as proposed is probably unworkable.

And also - this is the part most people miss - it doesn't matter whether the tax passes. What matters is that a serious person with serious political access has now put the idea into public circulation, and the direction of policy travel across every democracy facing rising youth unemployment is going to bend toward some version of it. Maybe not a token tax. Maybe disclosure requirements. Maybe mandated human-in-the-loop rules in regulated industries. Maybe tax credits for companies that keep human roles rather than replace them.

The founders who will be caught out are the ones whose entire pitch is "we are cheaper than a human." Their unit economics assume today's regulatory environment, and the regulatory environment is about to start moving.

The founders who will benefit are the ones whose pitch is "we make your existing humans dramatically more effective." The direction of political wind is now openly at their back.

Buried in the essay is a phrase Gates has coined and will probably keep using for the rest of his life: Human Reserved. His argument is that some categories of work will be formally set aside for humans, not because machines couldn't do them, but because societies will decide that having a machine do them would represent a loss.

His example is the caregivers who looked after his father through Alzheimer's. They understood his father even when his father couldn't express himself. A robot could probably do the physical work. It shouldn't. Something about the care, Gates writes, was irreplaceably human.

Whether or not Human Reserved ever becomes a formal legal category, it names a real emotional truth that quietly changes how a lot of markets are going to be built. There are things people want other people to do, regardless of what a machine can technically manage. Care. Teaching. Grief counselling. Certain forms of hospitality. Delivering hard news. Being the human on the other end of a decision that carries weight.

Every founder building anywhere near these categories now has a specific frame available that didn't exist a month ago. Your product is not "AI instead of a human." Your product is "AI so the human can do more, be more present, and cost the buyer less while staying in the room." That is a different sales conversation, a different premium, and a different regulatory posture.

The most interesting response to Gates's essay didn't come from Huang or Etzioni. It came from the MIT Technology Review interview Gates gave the same week. The interviewer asked how he would tell an AI that helps with invention apart from an AI that does job substitution, given that his tax proposal depends on that distinction.

Gates answered honestly. He said he doesn't know. He joked about Asimov's laws. He admitted the mechanism isn't there yet.

That admission is more important than any policy proposal in the essay. What Gates is really saying, underneath the tax and the Human Reserved framing and the calls for international governance, is that nobody - not Microsoft, not Nvidia, not the Gates Foundation, not the White House, not the EU - actually knows how to distinguish good AI from disruptive AI in real time. The people with the most information are openly saying they can't tell the difference, and they are asking for institutions that don't yet exist to help them work it out.

That is the future Gates is describing. Not a robot uprising. Not a utopia. A period of years in which the most powerful technology in human history spreads faster than any institution can respond to it, in which the pricing of human work gets rewritten sector by sector, in which the founders who move first get to write the rules of markets that don't exist yet, and in which the ones who wait for clarity get flattened by the ones who didn't.

If Gates is even half right, the next four years are going to reward a very specific kind of founder. Not the loudest. Not the best-funded. The ones paying attention to essays like this one and doing something about them before their competitors get around to reading the summary.

You just did the reading. Most of your competitors won't.

✅ Know a founder still building on the assumption that the next four years look like the last four? Forward this their way. The window between noticing and doing is where all the value gets made.

(👉 Vote now - we’ll share the results in next week’s issue. All votes are anonymous.)

(Last week’s poll results)

Access here and please share it with others.