AI put everyone back on
the same starting line.

What starting capital is actually for

I am not talking about multi-million businesses. I mean the general market: the stall, the workshop, the two-person agency, the online seller. Malaysia has 1.1 million micro, small and medium enterprises — 96.9% of all businesses — and about 70% of those are micro businesses.

For almost all of them, “capital” means one thing: the money to hire what you do not have. The skill you lack, the expertise you cannot do yourself, the experience you have not earned yet. A shop needs a designer, a marketer, someone to do the books. Rent and stock matter, but people are the line that decides whether you can start at all.

The Department of Statistics released its 2025 wage figures this week: skilled employees average RM5,799 a month. Hire three skills you lack at that rate — say design, marketing and accounts — and you are paying about RM17,400 a month before you have sold anything. That is what “wealth” has meant in the general market. Not luxury. The ability to buy skill.

Everyone is back at the same starting line

Now the same three skills are available, in first-draft form, for the price of a subscription. ChatGPT Go costs RM24 a month in Malaysia, and many tools cost nothing.

The person with RM17,400 a month and the person with RM24 are, for the first time, starting from roughly the same line on skill. The one with money can still hire people. The one without no longer has to, just to begin.

The woman supplying raw ingredients to hawker centres, who asked me what she had to buy before she could learn AI, does not need a marketing hire to write her price list in Malay, English and Chinese. I wrote about her in She Wants to Learn AI. I Told Her Not to Buy the Laptop Yet. I learned AI from zero myself, without hiring anyone to do it for me. The entry fee used to be a salary. Now it is curiosity.

China is already counting them

In China this has a name: the one-person company, or OPC. In the first half of 2025 alone, 2.86 million new one-person companies were registered — up 47% on a year earlier, and nearly one in four of all new businesses. By June 2025 there were more than 16 million of them. The report behind those figures is explicit about why: with AI models, agents and software tools combined, “one person can run the whole operation of a company.”

The legal form is not new. What is new is how many people now choose to start alone, because the skills they used to hire now come with the tools. Local governments have noticed: Hangzhou’s Shangcheng district has set aside a 1 billion yuan fund for them, and Shenzhen plans more than ten OPC communities by 2027.

A one-person company is the starting line made visible: a business that exists because its founder no longer needed the capital to hire a team first.

The evidence says the bottom gains most

This is not just my view. The best studies so far show AI narrowing gaps, not widening them.

A study of 5,172 customer-support agents, published in the Quarterly Journal of Economics, found an AI assistant raised productivity by 14% on average — 34% for novice and low-skilled workers, and minimal difference for the most experienced.

A Harvard and Boston Consulting Group experiment with 758 consultants found the same shape. Below-average performers improved by 43%. Above-average performers improved by 17%.

In both cases the people who gained most were the ones who started with the least. That is what a starting line looks like when it moves.

To be fair: the same line is not the same race

Two honest limits.

AI levels skill, not everything else. The person with money still has stock, a shopfront, a network, and months of runway if the first idea fails. Distribution and trust still cost money and time. China’s numbers show it: in one sample of one-person companies, only about 20% had found a business model that actually worked, and one person delivering alone tends to hit a ceiling of around RMB 1 to 1.2 million a year in revenue. The starting line is closer to equal. The race after it is not.

Skilled people do lose something. The experts gain least, so the gap between them and everyone else shrinks. What they lose is not the skill — it is the premium that existed because most people could not afford them, and the people paying it were exactly the small businesses in the general market. And skill still matters: in the same BCG study, on tasks beyond the AI’s ability, consultants using it were 19 percentage points less likely to get the right answer. Someone has to know when the machine is wrong. That judgment is what skilled people keep.

The other way round

Here is my version of the sentence: AI lets people without wealth get access to skills, and lets people with skills build wealth without capital.

Capital used to stand between an idea and a business, because it paid for the skills the founder did not have. It also stood between a skilled person and income, because skill needed an employer or a budget to turn into money. AI is shrinking both gaps.

What gets taken away is wealth’s head start: the automatic advantage of being able to pay for skill on day one.

What I would do with this

If you have an idea but no capital: start with what you have. Use the tools for the skills you would have hired, and spend real money only where AI cannot help — stock, trust, distribution.

If you have skill: package it. Sell the finished output, not your hours. The tools that multiply your work cost less than a phone plan.

If you employ people: pay for output, not years. I made that case in Job-Hopping Is Just Repricing. When everyone starts on the same line, output is the only fair measure.

The bottom line

In the general market, wealth mostly meant being able to hire the skills you did not have. AI did not hand that to the wealthy — they always had it. It handed a version of it to everyone else, and the data shows the people with the least gain the most. The race after the starting line still favours money. But the line itself has moved. It works the other way round — and a meme page with no designer just proved it.

If you see it the way the post does, I would like to hear why.