"The new arms race in computing power" is a May 22, 2026 long-form Financial Times feature by Charles Clover, the paper's security-and-defence correspondent. It argues that computing power is becoming a determinant of national sovereignty, examines the United Kingdom as a test case for middle-power compute strategy, and presents quantum and neuromorphic computing as technologies through which countries outside the US–China duopoly might secure an indispensable position in the supply chain.
Summary of argument
Clover writes that governments are beginning to treat compute "as earlier generations treated oil and electricity grids: as infrastructure on which economic strength, military capability and political independence depend," adding that "some even see computing power as a prerequisite for sovereignty itself, much as nuclear technology defined geopolitical power in the 20th century." He reports that the overwhelming majority of global AI computing power — "experts reckon 90 per cent" — is controlled by companies in the US and China.
The unit of analysis Clover uses is the "compute stack," which he defines as the "chips, networks, data centres and specialised hardware that underpin modern AI." Because no single country can replicate the whole stack, he describes a strategy in which a middle power aims to own one indispensable node: South Korea in memory (Samsung, SK Hynix), Taiwan in fabrication (TSMC), the Netherlands in EUV lithography (ASML), and, aspirationally, the UK in quantum, photonic networking, neuromorphic computing, and memory.
Quantitative anchors
| Datum | Value |
|---|---|
| Share of global AI compute controlled by US/China firms | ~90% |
| China quantum-program government spending | ~$17B |
| US quantum-program government spending | ~$9B |
| Japan quantum-program government spending | ~$9B |
| UK quantum funding initiative (March 2026 announcement) | £2B |
| UK "Sovereign AI" fund (April 2026) | £500M for homegrown AI start-ups |
| Trump UK visit deal (September 2025) — Nvidia/Microsoft commitments | >$100B |
| Universal Quantum German Space Agency contract | €67M quantum computer build |
| McKinsey estimate of AI data-centre electricity demand by 2030 | ~156 GW (≈1.5× Australia's installed capacity) |
| Oxford Ionics → IonQ (2025) acquisition | >$1B |
| US chip export controls (2022 framing — Clover paraphrase) | Forced China to make AI models 5–10× more efficient |
On national quantum spending, China leads at roughly $17 billion, with the US and Japan at roughly $9 billion each. Clover paraphrases the 2022 US chip export controls as having forced China to make AI models 5–10× more efficient.
Featured technical alternatives
Neuromorphic computing
Clover profiles brain-inspired computation at the UCL Neuroware Centre, directed by Tony Kenyon, which demonstrates neuromorphic approaches using event cameras. He reports that event-camera neuromorphic vision compresses cluttered scenes from "megabytes of data to kilobytes" by acting only when input changes, reducing bandwidth and power. Wetware variants, which integrate biological brain cells with silicon hardware, are referenced but are not the focus. André van Schaik (Manchester / Optera) frames the approach as "compatible with what we already use for microelectronics," positioning neuromorphic computing as augmenting rather than replacing GPUs. The UK Ministry of Defence has trialled Optera neuromorphic vision sensors for tracking satellites and space junk and has discussed drone tracking.
Quantum computing
The MoD bought a quantum computer from London's ORCA Computing in 2022, which Clover describes as probably the first quantum computer bought by a defence ministry. Sebastian Weidt, CEO of Universal Quantum, is quoted: "The UK, if one is honest, has probably lost the race in AI. In quantum computing, there is a unique opportunity, once in a generation, that we can build a trillion-dollar company here in the UK." Named military applications include the UK-Italian-Japanese Global Combat Air Programme fighter and encryption-breaking — for cryptocurrency and national secrets — once cryptanalytically-relevant quantum computers exist. Universal Quantum holds a €67 million German Space Agency contract to build a quantum computer.
Photonic networking
Oriole Networks, led by James Regan, is working with AMD on a system for the UK's Scaling Inference Lab. Clover frames photonic networking as a node where the UK could become the best supplier and thereby earn interdependence-leverage against US GPU export controls.
Specialty AI chips
Fractile, a London company, is reportedly in early-stage talks with Anthropic to supply specialized high-performance AI chips. Clover presents this as the first reported case of a major US frontier lab considering a UK specialty-chip supplier, within the wider account of model-makers diversifying away from Nvidia.
Geopolitical claims
Clover cites Trump's attempts to "prise Greenland from Denmark" earlier in 2026 as a reason US allies "have every right to fear that their access to US chips could be threatened," and frames this as a shift from prior coverage that assumed US export controls were stable. Dave Grimm of AlbionVC is quoted: "I wouldn't be surprised to see them use GPUs as leverage to some degree." James Regan of Oriole counters Jensen Huang's pitch that buying Nvidia GPUs provides sovereign AI with: "It is sovereign until a future American president decides that you can't have it." Nigel Toon, CEO of Graphcore, describes a scenario in which a foreign hyperscaler could weaponize dependence: "Let's say you were dependent on a large foreign hyperscaler for your compute. The way the security codes work, they could effectively brick the whole data centre wherever it was in the world."
The piece frames the central question of middle-power compute strategy through Boudewijn Wijnands of Fortaegis (Netherlands), who is quoted: "Right now a new generation of choke points is being created. You can call them choke points. You could call it interdependence." Clover treats this dual framing — whether dependence on US chips is leverage against (a choke point) or leverage with (interdependence) — as the operative analytical question.
UK domestic industrial-policy claims
Clover frames the UK's loss of DeepMind to Google, acquired in 2014, as the canonical case of British failure to convert scientific talent into globally competitive AI companies. He reports a March 2026 UK quantum funding initiative of £2 billion, an April 2026 "Sovereign AI" fund of £500 million for homegrown AI start-ups, and a September 2025 Trump UK-visit deal carrying more than $100 billion in Nvidia and Microsoft commitments. In a April 2026 speech, UK Technology Secretary Liz Kendall said she would not accept "defeatism" from those claiming the AI race had been lost. The article also notes a May 2026 pension-fund nudge toward private-market venture capital. Michael Cuthbert, director of the National Quantum Computing Centre, is quoted on retention: "There is a long-term ambition to do as much as the government can do to anchor UK start-ups and UK technology in the UK for UK impact." Clover separately notes the 2025 acquisition of Oxford Ionics by IonQ for more than $1 billion, and the McKinsey estimate that AI data-centre electricity demand will reach roughly 156 GW by 2030, about 1.5 times Australia's installed capacity.
Provenance
Charles Clover, "The new arms race in computing power," Financial Times, May 22, 2026 (security-and-defence desk). Raw text held at Raw Sources/The new arms race in computing power.md. It is the first major FT long-form treating compute sovereignty as a coherent policy frame in its own right rather than as a side-thread to US–China decoupling or semiconductor policy, and supplies a mid-2026 quantification of national quantum spending (China $17B vs. US/Japan $9B each).
Relationships
- supports: AI Sovereignty, Sovereign AI (Product Concept), Inference Economics and Token Pricing, AI and Cybersecurity, Export Controls (AI).
- related: Graphcore (Nigel Toon CEO, Bristol chip designer), Nvidia & TSMC — AI Compute Infrastructure, Anthropic (Fractile sourcing), China and the US Are Running Different AI Races, Military Ai, Autonomous cyber-agents, Quantum Computing Ai Implications.
- contradicts (partially): Sovereign AI (Product Concept) / Jensen Huang's "Nvidia = sovereign AI" pitch — Regan and Toon directly contest the framing.
- instance-of: Policy Essays On Compute Sovereignty.
Citation form
Cite as [[sources/clover-new-arms-race-computing-power]].