On 11 June the Guardian published a piece by Samantha Oltman with a diagnosis in its title: AI absolutism is breaking our brains. Her argument is that the two poles of the AI debate, boundless productivity on one side and the end of work on the other, are not opposing positions at all. They are the same sales pitch. Both tell you the technology's dominance is inevitable; both tell you to get on board or be left behind; both inflate the valuations of the people selling it. The day before, The Journal carried a piece by a senior policy analyst at the National Economic and Social Council, making the official Irish version of the argument: Ireland ranks fourth in the world for AI adoption, 45% of Irish people are using the tools, and we have a critical window to shape the technology through governance and literacy. Read together, the two pieces describe a debate. Read from Dublin, they describe a wager that has already been placed, with public money, by a State that talks about AI as a choice while budgeting for it as a certainty.

Oltman's case is worth setting out properly, because the evidence for it is stronger than a culture-war framing would suggest. The apocalyptic claims about AI and work have a documented author list, and the authors keep revising. Jensen Huang of Nvidia, 2025: "You're not going to lose your job to an AI, but you're going to lose your job to someone who uses AI." Dario Amodei of Anthropic, at Davos in January, described AI as not a substitute for specific human jobs but "a general labor substitute for humans". By March, venture capitalist Marc Andreessen was telling a podcast that large companies are overstaffed by 25 to 75% and that "now they all have the silver bullet excuse: ah, it's AI." By May, Sam Altman of OpenAI was conceding: "I thought there would have been more impact on entry-level white-collar jobs being eliminated by now than has actually happened", adding that he was "delighted to be wrong". Fortune's headline on that retreat did the analysis in one line: both men were walking back their jobs-apocalypse prophecies while eyeing blockbuster IPOs. Suresh Naidu, the Columbia economist Oltman quotes, explains the mechanism without conspiracy: "If you want to justify this enormous valuation in your IPO, you need to point to the revenue stream that you're going to generate in the future. You just need to make it look like you have something that can eat all the work on the planet."

The claims move with the funding cycle, not with the technology. The technology is real enough; the numbers underneath the panic check out and still support a more modest reading. Cumulative tech-industry layoffs since 2022 run to the high six figures by the trackers that exist, so Oltman's half a million is conservative, and economists she quotes attribute much of it to post-pandemic overhiring rather than machine replacement. AI capital expenditure now carries a startling share of American growth: estimates for the fourth quarter of 2025 range from roughly a fifth of year-on-year GDP growth at the cautious end to essentially all of the quarter's annualised growth at the other, depending on what you count. Naidu again, on the apocalypse itself: "Software is only about 4 to 6% of GDP. So it's a lot, but it's not like the whole economy can be replaced by Claude Code."

Readers of this site have seen the technology itself examined at length this spring, in Refusal Was an Option and Mirror, Mirror: what frontier capability is, who controls it and where it gets pointed. This piece is about the other half of the question, the half Oltman's framing opens and no British or American writer can finish. What happens when a state, rather than an investor, buys the inevitability story whole?

The genre, performed at state level

Start with the Journal piece, because it is a flawless specimen of the genre Oltman describes, performed in the register of Irish officialdom. The argument is careful, the tone is responsible, the prescriptions are governance frameworks and AI literacy. The headline facts are the tell. Ireland ranks fourth in the world for AI adoption; 45% of Irish people used AI tools in 2025. Trace those figures and you arrive at a single document: Microsoft's AI Diffusion Report, which ranks 147 countries by AI usage as measured through telemetry from Microsoft's own services. Ireland scores 44.6%, behind the UAE and Singapore, just behind Norway, just ahead of France. The fourth-in-the-world figure that anchors the official Irish account of AI adoption is the vendor's own usage data for the vendor's own products, published by the vendor.

The State's statisticians measure the same question differently. The CSO's Information Society survey for 2025 found that 20.2% of Irish enterprises use AI technologies in some capacity, up from 8.1% in 2023. That is precisely the EU average of 20.0%, less than half of Denmark's 42%. Among large enterprises the figure is 57.7%, which is where the Journal piece's "nearly 60%" lives. Both readings are arithmetically true, and the gap between them is the same gap this site keeps finding between Ireland's brochure and Ireland's accounts: world leader by the vendor's telemetry, dead average by the State's own survey. Official discourse cites the brochure.

The pattern goes to the top of the policy stack. Ireland's national AI strategy, published in 2021 and refreshed in November 2024, is titled AI: Here for Good. The pun is presumably intentional and it is doing more work than its authors noticed: for good meaning benefit, and for good meaning permanently, inevitability written into the title of the governing document. The refresh runs to seven strands, including one dedicated to AI infrastructure, advanced compute and data centre planning. It asks how Ireland can lead in adoption, how it can build trust, how it can skill the workforce. The one question it does not ask is Oltman's: who profits from the claim that all of this is inevitable, and what is the asking worth to them?

The ledger under the discourse

Here is what the official AI discourse never names. The Irish State is not a spectator to the AI buildout, or a regulator of it, or an adopter of it. It is a counterparty.

The numbers are familiar to readers of The Plan That Never Says Ireland and What the Surplus Hides, so the recital can be brief. Corporation tax receipts reached €32.9 billion in 2025, up 17.2% in a year, excluding the Apple money. The Irish Fiscal Advisory Council's February analysis found three corporate groups paying 46% of all corporation tax, around €13 billion, with two technology firms accounting for roughly 40%, around €11 billion. The Future Ireland Fund and the Infrastructure, Climate and Nature Fund, the State's twin sovereign funds, are capitalised from this windfall and held about €18 billion by February, managed in part by Amundi, BlackRock, State Street and UBS. The physical side of the position is larger still: data centres consumed 22% of all metered electricity in the State in 2024, up from 5% in 2015, more than all urban and rural households combined, with EirGrid projecting around 30% by the early 2030s. Dublin hosts the second-largest data centre cluster in Europe.

Put the discourse and the ledger side by side and the Irish version of AI absolutism comes into focus. For an American reader, the inevitability narrative is a story that investors buy. For the Irish State it is a fiscal strategy. The surplus is substantially an AI-era profits dividend; the sovereign fund is the dividend banked; the grid is the collateral physically posted. When Piketty said at the Global Justice Report launch that the billionaire class's new dream is "to cover the entire planet with datacentres", he was describing a global trajectory. There is one European country where the dream is furthest advanced, and its national AI strategy has a strand for accelerating it.

A state in that position does not need persuading that AI is inevitable. It needs AI to be inevitable. Which is why the official register sounds the way it sounds, and why the two endings of Oltman's absolutist binary deserve to be read here not as rival stories but as rival write-downs.

Ending one: the believers are right

Suppose the utopia pole pays out. The models keep improving, the revenue arrives, the data centre pipeline completes, the corporation tax keeps compounding. What does Ireland's win look like?

It looks like the grid surrendering a rising third of its output to the one project the State's climate law cannot absorb. The EPA's projections, published two weeks ago, show the second statutory carbon budget, 2026 to 2030, being exceeded by 53 to 82 megatonnes, a margin of more than a quarter of the budget itself, with national emissions falling at most 25% against a legislated 51%. The Fiscal Advisory Council and the Climate Change Advisory Council have jointly priced the EU compliance exposure at €8 billion to €26 billion, an estimate the responsible Minister dismissed as "back of the envelope stuff". Every data centre energised between now and 2030 widens that gap. In the ending where the believers are right, Ireland's reward is hosting the buildout that blows its own carbon budgets, paying a compliance bill that could exceed the annual health capital budget, and discovering that the labour transformation it was promised arrives mostly in the form Oltman identifies as the realistic one: not mass leisure but algorithmic management, the productivity squeeze already piloted on gig workers, working its way up the white-collar stack.

The win, in other words, is the scenario this site priced in The Plan That Never Says Ireland: rising material throughput, owned at the top of the wealth distribution, on a grid the State is failing to decarbonise. The believers' ending and the climate ledger are the same document read in different fonts.

Ending two: the sceptics are right

Now suppose the sceptics are right, the valuations are the IPO story Naidu describes, and the reassessment comes. The relevant warnings here are not from critics of the government. They are from the State's own advisory architecture, issued within the last four months.

The Fiscal Advisory Council, February, in the same report that counted the 46%: the technology firms carrying the tax base are making enormous AI investments that "may not deliver the high returns anticipated". The Central Bank, late May: Ireland is too reliant on corporation tax, the debt financing of AI in 2026 is already higher than in any previous full year, and any reassessment of the sector would have wider economic effects. The Fiscal Advisory Council again, yesterday: the corporation tax that was supposed to fill the sovereign funds is being spent, €5 of every €6 collected now going out through current expenditure, leaving the funds to be part-financed by borrowing. The May Exchequer returns already show the shape of it, a €2.3 billion deficit driven by the fund transfers themselves.

Assemble those three findings and the sceptics' ending prices itself. The tax base is concentrated in the firms making the AI bet. The windfall those firms generate has been committed to permanent spending. The fund that was supposed to hold the windfall against exactly this risk is being filled on credit. If the bubble deflates, Ireland does not merely lose a surplus; it loses the revenue stream that current expenditure has been built on, while holding a sovereign fund part-funded by debt and a planning pipeline of data centres whose owners have stopped building. The State's exposure to the AI story being false is not hypothetical or contrarian. It is the consensus position of its own fiscal watchdog and its own central bank, published, recent and unanswered.

The position, named

Oltman's piece ends on the observation that the version of AI we are being sold does not have to be the version we buy. For an individual that is advice about attention and tools. For Ireland it is a description of an open fiscal position. The State bought the inevitability story in the most literal way available to a state: it built the story into the revenue base, the expenditure base, the savings vehicle and the electricity grid, and it renews the purchase every budget day. AI absolutism is not breaking Ireland's brain. It is holding up Ireland's accounts.

Naidu offers the line that translates best. Asked about living through the AI rollout, he says an experiment implies a control group, and there is no control group. Ireland went further than most: it volunteered as the treatment group twice over, fiscally through the tax base and physically through the grid, and called the volunteering a strategy. The official discourse, the NESC register of critical windows and governance frameworks, performs deliberation above a position that has already been taken. That is the same pattern this site documented on the wealth question, where the State's silences and abstentions turned out to be its answer. Here the answer is louder. A country that titles its AI strategy Here for Good has told you what it has decided; the only live question is whether the deciding was done with anyone's consent.

Neither of Oltman's endings is inevitable. That is her point and it survives the crossing to Ireland intact. Inevitability is not a property of the technology; it is a property of the position, and positions can be exited. The State's own institutions have already drawn the exit map in fragments: the Fiscal Council's case for actually filling the funds, the Central Bank's case for widening the tax base, the EPA's arithmetic on what the grid can carry, the levers this site ran through the Irish case last week. What stands in the way is not the machine god and it is not the bubble. It is the convenience of a story in which nothing needs deciding because everything is already decided, told by people who profit from both of its endings, to a State that profits, for now, from believing them.

Sources

  1. Oltman, S., "AI absolutism is breaking our brains. The apocalyptic future we're being sold isn't inevitable", The Guardian, 11 June 2026. theguardian.com
  2. O'Sullivan, S., "AI and how we cope", The Journal, 10 June 2026. thejournal.ie
  3. Microsoft, AI Diffusion Report (H2 2025); Microsoft News Centre Europe, "Ireland Ranks 4th Globally for AI Adoption". news.microsoft.com
  4. CSO, Information Society Statistics, Enterprises 2025, Artificial Intelligence chapter, AI use by enterprise size and EU comparison. cso.ie
  5. Fortune, "Without AI spending, U.S. corporate capex would be negative, Pantheon analyst says", 23 February 2026. fortune.com
  6. Kedrosky, P., "Chart of the Day: AI Capex Ate the Economy in Q4", February 2026. paulkedrosky.com
  7. Federal Reserve Bank of St. Louis, "Tracking AI's Contribution to GDP Growth", January 2026. stlouisfed.org
  8. Layoffs.fyi tech layoff tracker. layoffs.fyi ; Crunchbase News layoffs tracker 2024-2026. news.crunchbase.com
  9. CNBC, "Anthropic CEO Dario Amodei warns AI may cause 'unusually painful' disruption to jobs", 27 January 2026 (Davos remarks, "general labor substitute"). cnbc.com
  10. Fortune, "Marc Andreessen says AI layoffs are a farce: Companies are 75% overstaffed and AI is the 'silver bullet excuse' to clean house", 31 March 2026 (20VC podcast remarks). fortune.com
  11. Euronews, "No AI 'jobs apocalypse' so far, says OpenAI's Sam Altman", 26 May 2026. euronews.com
  12. Fortune, "Sam Altman and Dario Amodei are both walking back their AI jobs apocalypse prophecies as they eye blockbuster IPOs", 26 May 2026. fortune.com
  13. Irish Fiscal Advisory Council, "More concentration, more risk", February 2026 (46% / three firms; AI returns caveat). fiscalcouncil.ie
  14. RTÉ, "Ireland too reliant on corporation tax - Central Bank", 27 May 2026. rte.ie
  15. RTÉ, "IFAC warns borrowing needed to part-finance saving funds", 10 June 2026 (Coffey: €5 of every €6 spent). rte.ie
  16. Department of Finance, end-2025 Exchequer returns, 6 January 2026. gov.ie
  17. RTÉ, Exchequer returns May 2026, 4 June 2026 (€2.3bn deficit driven by fund transfers). rte.ie
  18. NTMA, Future Ireland Fund and ICNF manager appointments, 3 February 2026. ntma.ie
  19. CSO, Data Centres Metered Electricity Consumption 2024, 10 June 2025. cso.ie
  20. EirGrid demand projections; reported at thejournal.ie
  21. EPA, greenhouse gas emissions projections 2025-2055, 26 May 2026. epa.ie
  22. IFAC and CCAC, "A Colossal Missed Opportunity", March 2025 (€8-26bn compliance exposure). climatecouncil.ie
  23. Department of Enterprise, Trade and Employment, AI - Here for Good: National Artificial Intelligence Strategy for Ireland (2021) and National AI Strategy Refresh 2024, 6 November 2024. enterprise.gov.ie ; gov.ie
  24. Chancel, L., Piketty, T., et al., The Global Justice Report, World Inequality Lab, 4 June 2026 (Piketty data centre remark at launch, via The Guardian, 4 June 2026). globaljusticeproject.wid.world ; theguardian.com