It costs $100/month to keep this site running — any contribution is appreciated.DONATE via Venmo @NoMoreDataCenters· Please share our links with like-minded people.

An Economist Told Australia's Parliament the Quiet Part: AI's Business Model Is Mass Unemployment — Not an Unintended Side Effect
BACK TO ARCHIVE
CRITICALCOMMUNITY IMPACT

An Economist Told Australia's Parliament the Quiet Part: AI's Business Model Is Mass Unemployment — Not an Unintended Side Effect

No More Data Centers EditorialOctober 20, 2026⌖ Australia — and the global AI data center buildout
SHARE

"Unlike any technology we've ever dealt with, the business model of AI is to generate mass unemployment. That is not an unintended consequence, it is the business model."
Which means every town that handed over its water, its grid, and its land for the promise of "data center jobs" signed up for the one thing the industry's own math says it will not deliver.

That sentence was not shouted at a protest. It was said, on the record, in a committee room in Canberra, by an economist.

Richard Denniss, co-CEO of The Australia Institute, appeared before Australia's Joint Select Committee on Artificial Intelligence — the parliamentary inquiry appointed on 20 August 2026, with submissions closing in September and a final report due 30 November. The 28-second clip his organisation posted afterwards has already been viewed tens of thousands of times, which tells you how rarely anyone in public life says the quiet part out loud.

"Unlike any technology we've ever dealt with, the business model of AI is to generate mass unemployment. That is not an unintended consequence, it is the business model. Now, as an economist, I understand productivity. People lose jobs when technology changes, but the valuations of these AI companies are literally based on the assumption that hundreds of millions of people will lose their jobs and employers will buy AI services instead of labor services."

Sit with the mechanism he is describing, because it is not a forecast of disruption. It is a balance sheet. The market value of the AI industry is not priced on AI doing new things — it is priced on AI doing existing work with the people removed. The buyers are employers. The product is the removal. Hundreds of millions of jobs are not the collateral damage of that trade. They are the revenue line.

Two pitches, and they can't both be true

Line them up side by side.

To investors: these models will replace human labour at a scale that justifies the valuations.

To host communities: this facility will bring jobs, growth, and a stronger tax base.

Both pitches are delivered by the same companies, to the same media cycle, using the same word. And the industry is perfectly clear about which one it means. It is why the jobs attached to a data centre project are almost always construction jobs — temporary, specialised crews who leave the week the concrete cures — while the permanent headcount that survives is a fraction of what the brochures implied. We have documented the pattern more than once: the "phantom workforce" of empty economic promises, the tax abatements that outlive the contractors who negotiated them, and the towns left holding the water bill and the humming.

A generator of mass unemployment does not owe a rural county 150 permanent jobs. On its own stated logic, it cannot provide them.

What Australia is being asked to pay while it waits

Australia is a useful case study because the buildout there is happening fast enough to watch in real time. Per the BBC's reporting:

  • Water. A single facility can consume up to 40 million litres of water a day — roughly what 80,000 households use. Sydney Water, the country's largest water utility, has warned data centres could consume up to 25% of Sydney's drinking water by 2035.
  • Power. The Australian Energy Market Operator projects data centre energy demand could triple nationwide by 2030, with backup diesel generators standing by for the moments the grid can't carry the load.
  • The green claims. Greenpeace Australia Pacific's analysis found no data centre operator it examined could adequately substantiate its claim of driving the country's renewable energy growth — and flagged a study commissioned by Amazon Web Services concluding that 6 in every 7 megawatts of grid connection requests are "phantom" demand.

So the bill is being paid in water, watts, land, and ratepayer dollars. The return, on the industry's own model, is fewer jobs. That is not a bad trade. It is not a trade at all.

In fairness: the labour market hasn't broken yet

It is worth saying plainly that the collapse the valuations assume has not shown up in the data. Yale's Budget Lab finds no discernible disruption in the broader labour market since generative AI arrived, and researchers at Stanford's SIEPR have pushed back on the loudest displacement claims, arguing the evidence for imminent mass job loss is thinner than the headlines suggest. Those are serious findings and they belong in this conversation.

They just don't rescue the pitch. A valuation built on a bet that hasn't paid off yet is still a bet — and host communities are being asked to front the infrastructure before the returns arrive. The industry keeps its optionality: it builds in case the bet wins, and keeps the local tax breaks either way. The town keeps a twenty-year water commitment and a fence.

The transparency ask is the whole ask

That is what Denniss was arguing for — not a verdict on whether AI is exciting, but transparency about who pays for the boom and who gets paid by it. It is the same question every planning board in America should be asking, and almost none of them can answer, because the economic impact studies placed in front of them are commissioned by the developers who stand to profit from the answer.

If the business model is replacing workers, then "jobs" is a brochure, not a contract. And if the numbers used to sell a project are inflated — phantom megawatts, temporary crews, five-year tax windows against thirty-year obligations — then the community is not a partner in the deal. It is the subsidy.

The quiet part is out. What happens next is just a matter of who repeats it, and how loudly, in the next county handed a rezoning request.

Say no to data centers.


Sources & Credits

Primary: "The business model of AI is to generate mass unemployment, it is not an unintended side effect," The Australia Institute — Richard Denniss, co-CEO, addressing the Joint Select Committee on Artificial Intelligence (Oct. 2026).
Inquiry: Parliament of Australia — Joint Select Committee on Artificial Intelligence (appointed 20 Aug 2026; submissions closed 14 Sep 2026; reporting 30 Nov 2026).
Economics: "Will the data centre 'boom' make us better off?" The Australia Institute; "What's the point of data centres?" — What's the Point? with Richard Denniss.
Australia's physical bill: "AI data centres are booming in Australia — but at what cost?" BBC News (40 million litres/day; 80,000 households; Sydney Water 25% of Sydney drinking water by 2035; AEMO tripling by 2030; backup diesel).
Energy & "phantom" demand: "Energy Vampires: The AI Data Centres Draining Australia," Greenpeace Australia Pacific (May 2026) — including Oxford Economics' finding, commissioned by Amazon Web Services, that 6 in 7 MW of connection requests are phantom demand.
Labour-market counterpoint: Yale Budget Lab — "Evaluating the Impact of AI on the Labor Market"; Stanford SIEPR — "What is really happening to jobs?"

// DID THIS RESONATE?

CLICK TO SHOW YOUR SUPPORT

DIALOGUE CHAMBER

(0 voices)
ESTIMATED ENERGY COST: 0.0000 kWh

NO MORE.DATA CENTERS

Exposing the physical weight of the digital cloud. Every byte has a cost. Every server has a footprint.

THE CLOUD IS NOT INVISIBLE

DATA HAS WEIGHT