The Hyperscaler Prisoner’s Dilemma: Why No One Can Afford to Spend Less

As part of results announcements in the last fortnight, the four dominant US ‘hyperscalers’ (Microsoft, Alphabet, Meta, and Amazon) have entered the largest capital spending cycle in the history of technology with a collective capex plan of ~$650 billion for 2026!

Alphabet stunned markets by projecting $175–$185 billion of capex for 2026, more than double 2025 levels.  Meta is guiding to $115–$135 billion, nearly double last year’s $72.2 billion.  Microsoft has not given a full year figure but is already spending $34.9–$37.5 billion per quarter on data centre build outs in FY 2026, far above any historical precedent (annualised this is $140-$150 billion). Amazon has just announced $200 billion in 2026 capex, equal to 143% of 2025 free cash flow.

Individually, each management team can justify its spend: demand for AI compute is surging, their services are supply constrained, and cloud backlogs are at record highs (Alphabet’s cloud backlog rose 55% sequentially to $240 billion; Microsoft’s RPO rose 51% to $392 billion), and customers are pushing for ever larger models, more chips, and more power-hungry workloads.  Collectively, however, these spending trajectories resemble something very close to a Prisoner’s Dilemma.

The Dilemma

The Prisoner’s Dilemma is a classic game‑theory thought experiment that illustrates how two perfectly rational individuals can make decisions that lead to a worse outcome for both, simply because they cannot trust the other to cooperate.  In the story, two suspects are arrested and questioned separately; each is offered the same deal: if you betray your partner while they stay silent, you go free and they receive a harsh sentence; if you both betray, you both receive a moderate sentence; and if you both stay silent, you each get only a minor charge.  The ‘rational’ choice for each prisoner, not knowing what the other will do, is to betray, because betrayal dominates silence in every scenario.  Yet when both follow this logic, they end up with the worst collective outcome, even though mutual cooperation would have left both markedly better off.

The hyperscaler equivalent looks like this:

• If one hyperscaler invests aggressively, it can win AI workloads, secure market share, and lock in multi-year cloud contracts.

• If a hyperscaler holds back, competitors will absorb that incremental demand, improve their models, strengthen their ecosystems, and compound their advantage.

• If all invest aggressively, capex balloons, returns compress, depreciation surges, and the entire industry shoulders the cost of an “AI arms race” that investors are already questioning.

In short: no one wants to spend this much, but no one dares not to.

The Choice

Cooperated (lower capex)

All four preserve margins, reduce pressure on their free cash flow, slow the depreciation overhang, but risk losing critical AI capacity in what is currently considered to be a ‘winner-takes-most’ market.

Defect (raise capex aggressively)

Captures AI workloads if others hesitate, but if all defect, everyone absorbs enormous capital intensity with little relative advantage.

Given Microsoft remains “capacity constrained through at least the end of the fiscal year” according to their CFO Amy Hood in November 2025 even after record spend, the dominant strategy, just like in the textbook dilemma, is to defect.

Why the Dilemma Persists

AI is supply constrained, not demand constrained

The limiting factor is compute capacity, not market appetite; whoever builds first wins the right to serve.

Scale advantages compound

Larger clusters improve utilisation, efficiency and training economics. Falling behind once can become permanent.

Customer lock in happens now, not later

Like most corporate tech projects, enterprise AI migrations will be multiyear undertakings. If a hyperscaler misses this wave, it cannot simply “catch up” later.

Market signalling matters

Investors punish high capex, but punishing lost AI relevance would be existential.

Alphabet’s Sundar Pichai summarised the entire dynamic unintentionally when asked what keeps leadership up at night: compute capacity.  If that is the binding constraint, then the rational move is to build as much of it as possible, no matter how much it costs.

A Classic Prisoner's Dilemma, but with Trillion-Dollar Stakes

The hyperscalers are functionally trapped in a strategic equilibrium where:

• Everyone defects (spends aggressively)

• Collective capex explodes

• No one can meaningfully slow down without risking long-term structural disadvantage

In game theory terms, hyperscaler capex has reached a Nash equilibrium: a stable but suboptimal outcome where deviation is too costly.

As with other Prisoner’s Dilemma scenarios, the logical decision for each player (invest more) is precisely what ensures the group spends far more than any would prefer in isolation.  A key point we have noted when listening to each hyperscaler’s earnings call is that when being asked the question about return on capital, the (paraphrased) answer from all four is “demand is sufficient to sustain growth”.  This is subtly different from a better answer which would start “we believe returns on this investment will be…”.  As any bank investor knows, being able to grow sales isn’t the same as being able to generate a return.

Conclusion

The hyperscalers are not irrational, but they may be constrained by the structure of the market in which they operate. Each must believe it can ultimately prosper, and some likely will, even if the collective outcome proves less attractive.

The Prisoner’s Dilemma is, of course, a thought experiment rather than a literal forecast.  Still, in the race to build global AI infrastructure, defection seems to have become the only viable strategy. As a result, 2026 is shaping up to be the most capital-intensive year in technology history, not because any one company desires it, but because none can afford to be the one that hesitates.

Risk Warning & Disclaimer

This article is provided for general information purposes only and is not intended to constitute investment advice, investment research, or a personal recommendation. It does not take into account the investment objectives, financial situation, or particular needs of any individual or entity.

The value of investments and any income derived from them may fall as well as rise, and investors may receive back less than they originally invested. Past performance is not a reliable indicator of future results. Where investments involve overseas assets, returns may be affected by movements in exchange rates.

The views expressed are those of Ptarmigan Capital as at the date of publication and are subject to change without notice. References to investment concepts, asset classes, or portfolio construction approaches are included for illustrative purposes only and should not be construed as a recommendation or a solicitation to buy or sell any security or financial instrument.

Ptarmigan Capital Limited is an employee-owned investment management firm providing discretionary investment management services to private clients, trusts, charities, and family offices. This article is intended to explain the firm’s perspective on bespoke investment management and the factors some clients consider when choosing such an approach. It does not relate to any specific investment product or strategy.

Ptarmigan Capital Limited is authorised and regulated by the Financial Conduct Authority (FRN 940407). Registered in England and Wales (Company No. 12715470). Registered office: 17 Cavendish Square, London W1G 0PH.