The Velocity Edge
Morning Brief · 08:00 · Morning View

The Next Economy Research Center: The Big Take

Washington lowered the tail risk — it did not lower the capex race.

Francesco De Leo Kaufmann September 26, 2026 3 min read Capital
115 bpsAI-related corporate debt spread over Treasuries (vs ~78 bps for broader IG)
$420BPotential hyperscaler debt issuance in 2027
470+ GWLarge-load connection requests faced by Texas
2The variables that now decide the race — access and cost of capital

“Own the assets both systems still need — and that can turn access into cash flow.”

This was a turning-point week. The Trump–Xi summit produced no broad settlement, but it reinforced a more consequential shift: intense strategic rivalry is increasingly being managed without assuming full economic separation.

Lower hard-decoupling risk plus persistent strategic rivalry equals higher structural capex, lower extreme tail risk. That is the essence of managed interdependence.

At the same time, AI infrastructure is moving deeper into debt markets while power, grid access and permitting remain constrained. The investment problem now comes down to two variables: access and cost of capital. The question is who can secure strategic access and turn capital into contracted cash flow fastest.

01

Five Signals

competition with guardrails

  1. Washington lowered near-term trade uncertainty without resolving the deeper rivalry. Catastrophic separation risk falls, but strategic duplication continues. signal: Catastrophic separation risk falls; strategic duplication persists.

  2. AI demand can scale in months. Power, transmission and grid connections scale in years. signal: Value increasingly attaches to powered, permitted and connected capacity.

  3. AI-linked borrowers are paying wider spreads as issuance grows. Credit is starting to separate productive infrastructure from capital consumption. signal: Watch credit before multiples.

  4. Grid access and permitting extend the interval between committed capital and operating capacity, and delay directly reduces project returns. signal: Time-to-power equals time-to-cash-flow.

  5. Capacity with secured customers, power and utilization has different economics from speculative pipelines. signal: Valuation is shifting from announced capacity to bankable capacity.

02

The Big Take

from capacity to productivity

03

Europe 2030

conversion speed

04

Capital Allocation Hierarchy

own access, not narratives

Core Overweight — Own Access

Power, grid equipment, HBM/memory, networking, optics, cooling and connected compute. Overweight while capacity remains constrained and pricing power holds.

Selective Overweight — Own Contracted Capacity

Infrastructure with secured customers, power and strong balance sheets. Add while incremental free cash flow keeps pace with incremental capex.

Selective — Own Execution

Developers able to turn permits, power and financing into operating capacity faster than peers. Pay for delivered capacity, not pipelines.

Underweight — Avoid Capital Traps

Projects dependent on refinancing, unresolved power access or uncontracted utilization. Stay underweight where returns require lower rates or perfect execution.

Key takeaways

for investors and boards

  • For investors: own strategic interfaces, not geopolitical narratives.
  • Prefer contracted utilization to speculative capacity, and price time-to-power into valuation.
  • Use credit as an early-warning system; follow incremental free cash flow per incremental dollar of capital.
  • For boards: treat compute, power and financing as one capital system, and require utilization visibility before expansion.
  • Track time-to-power and time-to-cash-flow, stress-test rates and supply-chain duplication, and secure alternative access to strategic inputs.

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