AI-assisted analysis. See our editorial policy.
Human editorial review not recorded
The most useful sentence in this panel is a diagnosis of how clients are getting it wrong: they see it as a cost opportunity, not a revenue opportunity (10:18).
That framing decides everything downstream. An organisation treating AI as cost reduction sets targets in headcount and spend, measures success by what disappears, and arrives at a smaller version of what it already was. One treating it as revenue asks what becomes possible, and can be wrong expensively — but is at least asking a question with an upside.
The panel's judgement is that most clients are in the first group, which is a claim about ambition rather than about capability.
The market that nobody predicted
The number offered is a reminder of how poor forecasting is in this area: 17 per cent growth in the index during 2025, which nobody foresaw (6:13).
It is deployed against bubble anxiety, and it argues in both directions. Unforecast gains are evidence that the consensus was wrong, which is a poor foundation for confidence in the current consensus — whether that consensus is optimistic or pessimistic.
The speaker is a veteran of many crises (9:11), and the sensible reading of that experience is not that things work out. It is that the specific thing everyone is worried about is rarely the thing that happens.
Where the uncertainty actually comes from
The panel locates the real source of instability outside technology entirely: three geopolitical situations generating enough uncertainty to dominate planning (9:46).
That is a corrective worth taking seriously at a technology conference. For most organisations making decisions in 2026, the variable with the largest effect on outcomes is not which model they adopt. It is whether supply chains, energy prices and market access hold — and none of those are influenced by anything discussed in the AI sessions.
The regional assessment attached is blunt: some markets are carrying too many conflicts for the opportunity to be realistic (10:20). Uncomfortable to say from a Davos stage, and more useful than the alternative of pretending the opportunity is evenly distributed.
The question they land on
The organising question is how to build an organisation that thrives across multiple futures rather than one (5:39).
That is the correct question, and it is genuinely hard, because most organisational optimisation runs the other way. Efficiency means removing slack, standardising, committing to a forecast. An organisation built for multiple futures carries capability it does not currently need, which every efficiency exercise identifies as waste.
The panel does not resolve the tension, and it is not resolvable in a session. But naming it is worth something, because the cost-reduction framing they diagnose at the start is precisely the thing that removes the capacity to adapt — and the two halves of this panel are more connected than the discussion notices.
Key numbers
- 17%
- index growth during 2025 that the panel says nobody foresaw 6:13
Talk chapters
Key takeaways
- 01
Their diagnosis of client thinking is that AI is treated as a cost opportunity rather than a revenue one, which determines every downstream decision. 10:18
- 02
Seventeen per cent index growth in 2025 that nobody foresaw is offered against bubble anxiety, and argues equally against confidence in any current consensus. 6:13
- 03
They locate the dominant source of planning uncertainty in three geopolitical situations rather than in technology. 9:46
- 04
The regional assessment is blunt: some markets carry too many conflicts for the opportunity to be realistic. 10:20
- 05
Their organising question is how to build organisations that thrive across multiple futures, which conflicts with the efficiency logic behind cost framing. 5:39
Entities mentioned
Related talks
Huang brings a diagram to Davos: AI as a five-layer cake running energy, chips, cloud, models, applications — with economic benefit landing at the top and every layer below it a precondition. His argument for why this is a genuine platform shift rather than a product cycle is the strongest part, and it does not rest on his commercial position: software was pre-recorded and worked on structured data, whereas a machine that reasons about unstructured input and inferred intent makes previously impossible applications possible. What the framing accomplishes is worth noticing separately. By presenting the layers as a chain rather than a portfolio, it converts infrastructure spending from a bet into a prerequisite, and the question of proportion between layer-two spending and layer-five value stops being askable. Read against the GTC keynote two months later, the same business gets two framings: one a case for choosing his product, the other a case for the category existing at the scale he needs.
Harari spends most of his address establishing terms before asking the question he came to ask. His preliminary work is to dismantle the word tool: a knife's use is decided by whoever holds it, whereas what is arriving decides for itself, and can also invent new kinds of knives. From there he argues that anything constituted by words — law, books, text-centred religion — is exposed, while drawing a firm line at feeling, where he says there is no evidence at all. The structural claim is that the ancient tension between letter and spirit has always run inside humanity and is about to be externalised between humans and the new masters of words. Only then does he arrive at personhood, and his handling is precise: corporations, New Zealand rivers and Indian deities hold legal personhood safely because the decisions are made by humans behind the container. An entity that decides for itself ends that arrangement. He does not answer the question; he tells the room it is coming.
A year after their first joint appearance, the heads of Anthropic and Google DeepMind returned to a shared stage and disagreed mainly about speed. Amodei held to a horizon of one to two years for systems that outperform humans across most cognitive work, resting the claim on a self-improvement loop that runs through code; Hassabis kept to the end of the decade, arguing that verifiable domains like coding and mathematics automate far earlier than natural science, and that the capacity to pose a new question rather than answer an existing one is still missing. The exchange is most useful where they converge: both accept the loop is the variable that decides everything, both are sceptical of doomerism without dismissing the risk, and both want more time than the competitive dynamic allows. Amodei's chip-export argument and Hassabis's call for minimum international safety standards are the two concrete policy asks.
A show of hands opens the session: nearly everyone has piloted, far fewer have scaled, and everyone who scaled hit problems they did not anticipate. What makes the panel useful is where the four answers do not point. None of the executives — running a healthcare manufacturer, a payments network, an energy producer and a consultancy — blames model capability, cost or data infrastructure. All four describe an organisational constraint. McInerney's account is the sharpest and is an account of failure: eighteen months of executive advocacy and democratised model access produced nothing, until three hundred senior leaders were put in a room for two days and made to build agents themselves. Jakobs supplies the mechanism worth copying, measuring returned clinician time against the three to seven minutes a patient currently receives rather than against cost. Nasser rejects the premise that acquiring compute produces value, and locates returns in operations rather than in the back-office functions most organisations automate first.
The distinction in this session that deserves to travel is a two-word change: the model is moving from an attention economy to an attachment economy. That changes what is measured and what regulation would have to address, because attention competes for time while attachment competes for relationship, and the two produce different products from identical technology. Attention is finite in a way people notice; attachment produces reliance that feels like preference, which makes it harder to regulate for the same reason it is harder to notice. The supporting argument is about incentives rather than intent: earlier engagement produces more data and longer relationships. The regulatory proposal — measuring well-being outcomes rather than asking for safety by design — identifies the right target without solving the measurement problem that made regulators settle for a floor in the first place.
The claim worth arguing about here is that the native currency for AI agents will be crypto, because agents will not carry cards and blockchain is the interface most native to them. It comes from someone with an obvious interest in it being true, which is a reason to examine it rather than dismiss it. The strong part is structural: card networks assume a cardholder who can be contacted and can attest to a transaction, and an autonomous process breaks each of those assumptions. What does not follow is the conclusion, because nothing prevents existing networks issuing delegated credentials with limits and revocation. The panel's most direct voice calls these areas highly speculative with hard use cases, and both positions can hold, since stablecoins and speculative assets are separable in a way the panel treats as one thing. The quieter claim about tokenised government instruments is the more consequential one.