THINK Ahead: How to survive without AI
Per the full note from ING's James Smith, the desk is arguing that the global economy's growing dependence on AI infrastructure creates a concentrated fragility that markets are underpricing — and that a 2027 'No-AI' scenario is a fatter tail than the 2% of webinar attendees who flagged supply shocks would suggest. The evidence is the bank's own audience poll: just 2% picked new supply shocks including cyberattacks as the top 2027 risk, versus 34% who chose an AI bubble bursting. Smith's framing is that a chatbot outage is trivial, but the embeddedness of AI in power, payments and compute is not — and that the damage function from critical-infrastructure failure is convex. The piece is a narrative risk note, not a forecast, so there is no FX pair or policy target attached; it sits behind the AI-capex channel that currently drives US equity and rate expectations.
What the desk is arguing
The thesis, stated plainly: the market is underweight the resilience risk embedded in AI dependence, and overweights the bubble-bursting risk by comparison. Per the full note , Smith uses a same-day Copilot/Microsoft outage as the hook, then pivots to the harder question of whether the global economy could function in 2027 without AI. The desk's implicit claim is that AI is now critical infrastructure, and critical infrastructure failures are underpriced.
The supporting evidence is the webinar poll split: only 2% of the audience chose new supply shocks (including cyberattacks) as the biggest 2027 risk, against 34% for an AI bubble bursting and stock markets crashing. Smith leans on the Spain blackout of 2025 as a precedent — it barely registered in growth figures, which cuts both ways: outages are survivable, but the tail is about duration and contagion, not a few hours. The note is explicitly a resilience argument rather than a growth forecast.
The counterfactual the desk is rejecting is the consensus framing that AI risk is purely a valuation/bubble risk. If AI is infrastructure, the risk is operational and correlated — a single vendor outage propagates through payments, cloud and power. That is a different, and less hedged, exposure than an equity drawdown.
How other firms see it
The related macro channels are the AI-capex cycle and its funding: US technology capex, the US equity complex, and the rates curve that discounts it. Watch USD/JPY and the US 10-year as the transmission vehicles — a resilience shock would hit risk assets and the dollar together, while a bubble-burst would be a classic risk-off dollar bid. EUR/USD is the cleaner read on the European growth spillover if AI capex slows.
What the calendar says
No high-impact events sit on the calendar in the next 30 days for this jurisdiction, so there is no scheduled catalyst to trade against. Treat this as a slow-burn narrative risk rather than an event-driven trade; the tell will be in positioning and in AI-capex guidance from the next earnings cycle, not in a data print.
Key takeaways
- 01ING's James Smith frames AI as critical infrastructure, making resilience risk — not just bubble risk — the underpriced tail for 2027.
- 02Only 2% of ING's Global Outlook webinar audience picked supply shocks/cyberattacks as the top 2027 risk; 34% chose an AI bubble bursting.
- 03The Spain 2025 blackout is the desk's precedent: short outages are survivable, but duration and contagion are the real damage channel.
- 04No FX pair or policy target is attached — the read-through is via AI capex, US equities and the rates curve.
Market implications
Watch US technology capex guidance and the US 10-year yield as the primary transmission of an AI-resilience shock; a genuine infrastructure failure would hit risk assets and the dollar simultaneously, unlike a bubble-burst which is a risk-off dollar bid. EUR/USD is the cleanest European spillover read if AI capex slows. Positioning in AI-linked equities is the signal to monitor, since the note is a narrative-risk argument rather than a data-driven call.
Risks to this view
The call is invalidated if AI capex guidance holds up through the next earnings cycle and no infrastructure incident materialises — the resilience tail then stays unpriced and irrelevant. A genuine AI bubble-burst, the 34% scenario, would dominate the narrative and make the resilience question academic. Any real cyberattack or extended cloud outage would force a fast repricing of the 2% tail, cutting the other way.
Opinions Opinion by James Smith THINK Ahead: How to survive without AI Published 14:55 Translate We've all had to learn to live without a *certain* chatbot for the last 24 hours, but could the global economy survive without AI in 2027? James Smith goes all agentic and looks at how the economic effects of AI could change over the coming months The global economy without AI? Let me ask Copilot Seriously, what are we supposed to do when an AI chatbot breaks down?
Who’s going to tell me I’ve made a “really insightful point” after a seven-word prompt, with three words spelled incorrectly? Who’s going to suggest a clickbait headline, like “Unpacking a world without Copilot: James Smith’s THINK Ahead at a crossroads” ? And, most importantly, who’s going to ask whether I’d like the answer turned into a table?
After briefly confronting the terrifying prospect of thinking for ourselves, I thought I’d have a go at doing AI’s job. So, after today’s outage, here’s a timely prompt: Can the global economy survive without AI in 2027? Connecting the dots… Piecing it all together… Pretending to think deeply… That’s a fascinating question! (Predictable.) Let’s break it down into four key areas.
Because everything has to be broken down into four key areas… The resilience question If today’s Microsoft outage tells us anything, it’s that we probably underestimate the economic risks of critical infrastructure failing. At our Global Outlook webinar this week, just 2% of the audience picked new supply shocks, including cyberattacks, as the biggest risk for 2027. A rather larger 34% chose an AI bubble bursting and stock markets crashing.
More on that shortly. Now, one chatbot crashing out for a few hours isn’t going to move global GDP. Spain’s huge blackout in 2025 barely registered in the growth figures.
But as we argued in our outlook , the damage from electricity, payments or communications outages could rise disproportionately if they become more frequent or widespread. It starts to hit confidence. And businesses are forced to spend more insulating themselves.
One data centre owner told me this week that they keep 48 hours’ worth of diesel on site in case the power conks out. That sort of contingency doesn’t come cheap. But the flipside is that businesses are building up resilience against supply shocks like the Iran war – or a ‘moat’, as our AI overlords like to say.
Greater preparedness, including more hedging, is one leading explanation for why European core inflation has remained remarkably subdued so far. What is the biggest risk in 2027? (Our webinar poll) Based on 202 votes Source: ING "> Based on 202 votes Source: ING ‘Unpacking’ (sorry) the growth impact Everyone talks about the boost AI investment has given the US economy. And my colleague James Knightley wrote in our outlook this week that this is likely to continue into 2027.
What gets talked about less is the impact it’s having on Europe. Partly, that’s because the story is mixed. As a nice chart from my colleague Pablo Muylle showed this week, not everyone is enjoying a Netherlands/ASML-style boost to manufacturing.
It’s a similar story for data centres, which are contributing around 0.1-0.2pp to year-on-year eurozone growth if you take the IT sector as a proxy. By contrast, as we saw in this week's GDP data , the data centre rollout in the UK is adding around 0.4pps to the roughly 2% annual growth we’re seeing. That should remain a tailwind next year and there’s clearly scope for a positive surprise.
The more intriguing question is what’s happening outside the server farms. UK output across parts of professional services, from law to accountancy, has shot up remarkably quickly this year and in apparent unison. Is that AI productivity coming through?
I suspect that is some, though not all, of what’s going on. Or as AI would put it, “The answer is nuanced.” Thanks for that… Is this the first sign of AI productivity showing up? Source: Macrobond, ING "> Source: Macrobond, ING What if the AI bubble bursts?
Never ask an AI chatbot if AI is a bubble. It hurts our feelings. But… if we did get a proper bear market in US equities, the problems are likely to extend well beyond tech.
We wrote about this in our Outlook report this week. Remember what happened after the dotcom bust, when non-tech investment fell harder than tech itself, on account of wider credit spreads and tighter lending standards. And were it to happen – not our base case – then it would have big ramifications for the Fed, which has built a rate hike rationale linked heavily to the solid equity market.
My FX colleagues reckon the impact on the dollar would depend not only on that interest rate story, but whether an equity correction threatens financial stability. If it does, the dollar could still attract some safe haven flows, though in general they reckon JPY and CHF are the more obvious beneficiaries of a serious sell-off. Read more in their FX Outlook hot off the press this week – and sign up for the team's webinar on Monday. 'Chipflation' is not a big thing right now Source: Macrobond, ING "> Source: Macrobond, ING And what about inflation?
Fed Governor Chris Waller suggested this week that AI is putting upward pressure on inflation. It helped form the basis of his suggestion that rates could rise by 50bp in December, and contributed to the 32bp rise in short-dated US rates this week. But is it true?
There are admittedly many ways AI could be influencing inflation, but the evidence of “chipflation” is pretty flimsy so far. Even using a generous definition of electronics and IT hardware, those categories are still dragging US inflation down, not up. So are cars, despite being stuffed with semiconductors.
The inflation figures are undoubtedly problematic; a 0.4pp rise in the Fed’s preferred metric of core inflation can’t be ruled out for October. But AI really isn’t the main event here. So there you have it.
Now, would you like me to summarise that in five bullet points, adopt a more professional tone, or turn it into a PowerPoint presentation featuring an alarming amount of purple? James Smith THINK Ahead in developed markets United States (James Knightley) Financial markets have aggressively repriced the outlook for Federal Reserve policy in the wake of hawkish comments from officials and a hotter-than-anticipated core CPI print. There are several Fed officials scheduled to speak over the coming week, despite Chair Kevin Warsh's desire for less forward guidance, and they will retain a hawkish bias.
We fully expect a 25bp rate hike at the December FOMC meeting, but the Fed will have a pre-release copy of the November inflation report, and that should look very benign given steep declines in gasoline prices. Hence, we don't see much likelihood of a mooted 50bp hike. Retail Sales/Industrial Output (Tue): In terms of data, we will have October retail sales, which will be held down by weak auto sales.
Gasoline sales are likely to be close to flat. In real terms, we could see total sales fall. Manufacturing output should continue growing, boosted by a strong performance from advanced manufacturing tied to the tech/AI rollout.
Some traditional, heavy energy usage sectors, such as steel and glass, should also perform well given their comparative advantage over European and Asian producers, who are having to deal with sharply higher energy costs. United Kingdom (James Smith) Inflation (Wed): We expect headline inflation to nudge higher on a small rise in the quarterly energy price cap and a further 4% rise in fuel costs. It won't be until January that we see the full impact of higher natural gas prices when household energy bills are set to rise by circa 25%, something that's likely to be formally confirmed in the coming days.
That's likely to take headline CPI to just shy of 4.5% early next year. However, for now, second-round effects look minimal. We don't expect further signs of that in October's data; food inflation is likely to stay unchanged in year-on-year terms.
UK-EU summit (Fri): A deal is widely expected on 20 November to bring the UK closer to Europe, via agreements to reduce checks/paperwork on agricultural goods, align on emissions trading, and increase youth mobility. We'll be looking for any clues about PM Burnham's willingness to go further, and what these deals tell us about his willingness to compromise on key issues of migration and regulatory alignment. That will hint at how close he might be prepared to go to the single market in future.
Key events in developed markets Source: Refinitiv, ING "> Source: Refinitiv, ING Content Disclaimer This publication has been prepared by ING solely for information purposes irrespective of a particular user's means, financial situation or investment objectives. The information does not constitute investment recommendation, and nor is it investment, legal or tax advice or an offer or solicitation to purchase or sell any financial instrument. Read more THINK Ahead Share X LinkedIn E-mail Copy link Share X LinkedIn E-mail Copy link Download Share X LinkedIn E-mail Copy link Share X LinkedIn E-mail Copy link Download In this opinion The resilience question ‘Unpacking’ (sorry) the growth impact What if the AI bubble bursts?
And what about inflation? THINK Ahead in developed markets Author James Smith Developed Markets Economist, UK James is a developed market economist, responsible for ING's view on the UK economy and Bank of England. He graduated from the University of Bath with a degree in economics and joined ING in 2015.
Sources & References
How we cover this story
Related news on this pair
EUR/USD: Support holds but upside limited – ING
EUR/USD support at 1.1110 likely temporary; weak bounce momentum suggests USD strength remains favored on any retracement.
Euro: Hawkish ECB cushions losses against US Dollar - ING
ECB hawkish positioning provides near-term floor for EUR/USD around 1.1200 despite dollar strength from potential US political outcomes.