The Velocity Edge
GVI · 08:00 · GVI Global Cities Signature Survey

Global Cities Ranking

AI infrastructure is no longer constrained only by chips and power — cities are beginning to price the social licence to compute.

Francesco De Leo Kaufmann September 19, 2026 7 min read Urban Velocity
$1M/MWProposed New York community contribution per MW of new data-centre demand
$71.8BGlobal H1 cross-border commercial-property investment, +56% YoY
$8.7BSingapore's cross-border property volume — highest globally
25,000Lucid/Bolt Level-4 autonomous vehicles planned for European cities

“The next scarce resource is not power. It is permission.”

The most consequential urban signal this week is a new one: permission is becoming a factor of production. New York is recommending community-investment requirements of at least $1 million for every megawatt of new data-centre utility demand after the state imposed a moratorium on large new facilities. Virginia — the world's most concentrated data-centre geography — is tightening permitting, transparency and environmental controls. The implication is bigger than regulation: communities are beginning to demand a direct economic return for surrendering electricity, water, land and infrastructure capacity to AI.

At the same time, capital continues to concentrate in urban ecosystems capable of absorbing it. Global cross-border commercial-property investment jumped 56% to $71.8 billion in H1 2026; Singapore ranked first globally with $8.7 billion, while London and Milan were among the European cities attracting particularly strong international capital. Meanwhile, Taipei strengthens its position in the physical AI stack as MediaTek launched its first smartphone processor on TSMC's 2-nanometre process and expands into data-centre accelerators after a $3.9 billion convertible financing backed by Nvidia and Alphabet.

And AI is crossing another boundary — from digital intelligence into the physical economy. Anthropic has established a wet biology lab in the San Francisco Bay Area to combine AI with robotic experimentation, while Tallinn-headquartered Bolt and Lucid plan to deploy 25,000 Level-4 autonomous vehicles across major European cities, within a broader ambition for a 100,000-vehicle autonomous fleet by 2035.

01

The GVI Top of Table

Where the structural anchors hold

02

The Movers

Cities gaining and losing ground

  1. London gains 0.3 points to 79.7. International property investors are returning strongly to premium London assets, while the discovery layer keeps generating companies: London-based Mantic raised $25 million — backed by Microsoft's M12 and Balderton Capital — after its forecasting AI outperformed human contestants in a major 2026 competition. signal: The persistence of the London mechanism — DeepMind alumni → frontier capability → venture formation → institutional demand → global capital.

  2. San Francisco rises to 79.5, not because its housing problem has disappeared — it has not — but because the ecosystem is moving into a new layer of the economy. Anthropic's Bay Area wet lab combines AI models, life sciences and physical automation, letting software intelligence direct robotic scientific experimentation. signal: The transition from AI-native software to Industrial AI embedded in the GVI theory.

  3. Taipei rises to #27 (61.9). MediaTek's 2-nm Dimensity 9600 Pro carries a neural processor claimed to improve prompt processing by 51%, while the company moves into data-centre accelerators and custom silicon. Its $3.9 billion convertible financing included $3.5 billion from Nvidia, with Alphabet also participating. signal: Semiconductors → edge AI → custom accelerators → cloud integration → higher hardware-system value.

  4. Hong Kong moves to #35 (58.8) and records the largest Top-50 score improvement alongside Tallinn, both +0.5. RoboTechnik, a maker of silicon-photonics equipment for data centres, plans a further $800 million secondary listing — reinforcing Hong Kong's role as a financing bridge between Chinese industrial-AI companies and international capital. signal: Capital-market machinery is reconnecting with the AI supply chain.

  5. Milan gains 0.4 points to 56.5 without yet moving position. International real-estate investors have become particularly active in Milan alongside London as European cross-border capital returned sharply in H1 — validating a thesis long dependent on converting industrial depth and Northern Italian infrastructure into international capital formation. signal: International institutional capital strengthens the Milan optionality.

03

Urban Velocity Leaders

Functional specialization emerges

Singapore

Global capital × execution × low institutional friction. Singapore orchestrates the complete system.

Seoul

Chips × exports × AI reinvestment. Seoul and Taipei industrialize intelligence.

Hong Kong

IPO liquidity × AI-infrastructure financing. Hong Kong and London finance it.

Paris

Frontier AI × sovereign/private capital. Paris and San Francisco create intelligence.

Helsinki

Abundant clean power × accelerated computing. Helsinki powers it.

Taipei

2-nm silicon × edge AI × custom compute — the physical AI stack.

04

Great Cities at Risk

When abundance meets backlash

05

Cities to Watch

Beyond the Top 50

Five key signals

The Velocity Age is specializing

  • Social licence becomes infrastructure. New York and Virginia show the race entering a political-economic phase; power, land and water are not enough — developers need community legitimacy.
  • Global capital is returning to metropolitan concentration. Cross-border property investment rose 56% in H1, with Singapore first globally and London and Milan particularly active.
  • AI is moving from the screen into the laboratory. Anthropic's Bay Area wet lab signals that the next urban cluster may need robotics and life-science infrastructure alongside GPU capacity.
  • Autonomous mobility is crossing from demonstration to industrial deployment. Bolt and Lucid's 25,000-vehicle European plan shifts Level-4 from pilot to scale.
  • The chip race is spreading from training to ubiquitous inference. MediaTek's 2-nm platform, its cloud accelerators and Axelera's enterprise push broaden compute from giant clusters to servers, devices, factories and mobility.

Francesco De Leo Kaufmann · The Velocity Edge

More from The Velocity Edge

Jump straight into another note — no need to head back.

Keep reading

More field notes, from dawn to close.

Day signals and night briefings, published like a journal for board-level AI decisions.

Back to The Velocity Edge