The house isn’t the risk.
The thirty years are.

A video came across my feed this week arguing about whether young Malaysians should buy a house or keep renting. Someone had bought at eighteen and lost RM100,000. The comments divided into the two camps these debates always produce, and both of them were arguing about the wrong thing.

Buy or rent is the wrong question

Every one of these debates runs on the same axis: is property a good investment, or is rent money thrown away? Both sides then argue about price direction, which nobody can forecast.

Change the axis. A mortgage is not a purchase — it is a thirty-year commitment of income you have not earned yet. The house is collateral. The actual product being sold is a promise: that for the next three decades, roughly a third of your monthly income will arrive on schedule, from work that still exists, in a currency that still buys what it buys now.

That is the thing to examine. Not the price of the unit. The confidence level on the promise.

We have seen this structure fail already

I have made this argument before, about a different asset. Education used to be a prepayment system — sixteen years of study paid upfront, against forty years of returns. It worked while knowledge depreciated more slowly than a career. Then the half-life of skills fell below the length of a degree, and the safest degree fell first.

A mortgage is the same structure with a longer tail. You prepay three decades of income certainty against an asset whose value you will discover slowly. The structure only holds if the two clocks match — if your income is at least as predictable as the loan is long.

That match is what broke. Not the property market. The income clock.

The income side stopped being predictable

Look at what we can already measure. Malaysia recorded 31,082 retrenchments in the first four months of 2026 — and as I argued at the time, most of that had nothing to do with AI yet, which is precisely what should worry us.

Meanwhile the way pay itself works has changed. Staying put earns roughly 5% a year; moving reprices you 15–20%. Careers are now priced in three-year cycles, not thirty-year arcs. Ask a 26-year-old to describe their job in 2040 and they cannot — not because they lack ambition, but because the honest answer is unknowable in a way it was not for their parents.

So the bank asks for thirty years of certainty, and the economy now issues it in three-year blocks. Nobody is pricing that mismatch, because the mismatch does not appear on the loan document.

Ask China’s film industry how long it took

Before anyone tells me disruption is slow and generational, look at what just happened to an entire industry in China — not a forecast, a completed quarter.

In the first quarter of 2026, China had roughly 128,000 micro-dramas online. About 122,000 of them were AI-generated — around 95%. Three years ago AI video could not hold a human face steady for four seconds. It is now the default production method for an entire content category.

The human numbers are worse than the market-share number. Short-drama actors who commanded day rates around RMB 20,000 now get about RMB 1,200 — and still struggle to find work. Production teams that ran 50 people now run 5. Ninety per cent of the roles on a production, gone — not relocated, not retrained, gone.

And the detail that tells you it is not reversible: a 29-year-old former train driver produced an original short film in ten days for about RMB 3,000, alone, with a current-generation model. Caixin ran the whole story as a cover report — AI moving through the entire film and television value chain, not one department of it.

That did not take a generation. It took about three years, and the steepest part took one.

I am not watching this from outside. Producing AI video commercially is part of what my company sells, so I see this curve from inside the workflow — and it moves faster than anything I have worked with.

Twenty-four months of runway

So let me say what I am doing rather than only what I think.

AI capability is now measured in months. Not product cycles, not annual roadmaps — months. A model that reset our cost assumptions in one quarter is mid-tier by the next. That pace is exciting when you are building on it. It is something else entirely when you are forecasting against it.

So at IMA we are raising our backup cash flow to twenty-four months. Not because business is bad, and not because we expect trouble — but because when the ground moves this fast, the only honest response to uncertainty is to buy time. Runway is not pessimism. It is the price of being able to change your mind.

Now put the two numbers beside each other. A company operating directly inside this technology, with revenue and clients and every reason to be optimistic, is extending its margin of safety to twenty-four months — because twenty-four months is as far as I can honestly see.

A mortgage asks for three hundred and sixty.

We measure our safety in months. The loan is denominated in decades. I am not telling anyone to refuse one. I am saying I notice the gap, and I have already decided what to do about it on my side of the table.

The demand side is already counted

Now the part almost nobody in these videos raises, and the reason I think this is structural rather than cyclical.

Malaysia’s total fertility rate is 1.6 — below the 2.1 replacement level, down from 1.7 the year before. In 2024 the country recorded 414,918 live births, the fewest since 1980, a 9% fall in a single year. The decline continued into 2026.

Property is a bet on future demand. The buyers for the unit you purchase today are, in large part, people who must first be born. The generation that would buy your house in 2050 is being counted right now, and it is smaller every quarter. That does not crash prices next year. But anyone signing a thirty-year commitment is, by definition, underwriting exactly that horizon.

And the supply side is flashing

The near-term picture is not tight either. In Q1 2026 Malaysia held 32,801 unsold completed homes worth RM16.37 billion — the sixth consecutive quarterly increase, and up 39.5% from a year earlier. Add 19,263 unsold completed serviced apartments worth RM16.52 billion, nearly ten thousand of them in Johor alone.

One detail deserves attention. Unsold volume rose 7.6% in the quarter while the total value of that overhang fell 7.7%. The stock is piling up at the cheaper end, and prices are being trimmed to move it — which is precisely the segment first-time buyers shop in.

Set that against affordability: a median house around RM494,000 against median household income of RM8,479 a month — roughly 4.9 times annual income, where the international threshold for affordable is 3. Malaysia has been above 4 since 2002, and researchers have called housing here seriously unaffordable for years.

To be fair — the honest case for buying

I would be doing you a disservice if I argued only one side, so here is the other, properly.

Rent is not free. You are paying somebody’s mortgage either way, and rent rises with inflation while a fixed instalment does not. Over thirty years that gap compounds in the owner’s favour, and it is the single strongest argument for buying early.

Location still separates outcomes. The overhang is concentrated — serviced apartments, specific states, specific price bands. Landed property in mature areas with real transport access has not behaved like the glut, and averaging them together hides more than it reveals.

A home is not only an asset. Security, stability, a place children grow up in, freedom from a landlord’s decisions — these are real returns that never appear in a yield calculation. Many people who “overpaid” on the spreadsheet made the right decision for their life.

And demographics move slowly. A 1.6 fertility rate reshapes 2050, not 2028, and immigration can offset a great deal. Anyone claiming to know the 2050 price of your unit — in either direction — is guessing.

One more thing, plainly: I am a technology CEO, not a financial adviser. What I have written is how I reason about risk, not a recommendation about your money.

The bottom line

Rather than buy or rent, I would ask three questions before signing anything.

How many years of income am I committing, and how honestly can I describe year ten? If the answer is “I assume it continues,” that is an assumption, not a plan.

What happens if my income halves for twelve months? Not zero — halved, which is the common case in a repricing market. If the answer involves selling the asset into a soft market, the commitment is larger than the affordability calculator suggested.

Am I buying a home or an investment? They are different decisions wearing the same paperwork. If it is a home you intend to live in for twenty years and you can carry it through a bad year, most of this article matters less to you. If it is an investment premised on someone paying more later — look again at the birth data.

Flexibility used to be the expensive option and commitment the safe one. In a job market that reprices every three years, that relationship may have quietly inverted. That is the part worth thinking about before the paperwork, not after.