The Velocity Edge
Morning Brief · 08:05

The Next Economy Research Centre

The IMF's October outlook lifts 2026 global growth to 3.1% even as it warns that the AI investment cycle is becoming a systemic financial risk.

Francesco De Leo Kaufmann October 13, 2026 4 min read Macro
3.1%projected global GDP growth in 2026, supported by AI investment
10%share of global investment-grade bond issuance now from hyperscalers (vs <1% in 2024)
32countries accounting for 95% of global growth, down from 54 in April

The International Monetary Fund has delivered its most consequential reading yet of the AI-native economy. Its October 13 World Economic Outlook raises projected global growth for 2026 to 3.1%, from 3.0% in July, despite the energy shock tied to the Middle East conflict. Yet its companion Global Financial Stability Report warns that the same AI investment cycle supporting growth could become a source of financial instability.

AI is supporting aggregate growth; its financing is becoming central to global credit markets; and the geographical distribution of its benefits is narrowing. The emerging risk is that AI becomes simultaneously a source of macroeconomic resilience and a transmitter of financial instability.

This strengthens the Great Split thesis developed by NER-C. The distinction is no longer between economies that invest in AI and those that do not, but between those capable of transforming AI investment into productive capacity and those accumulating financial exposure without equivalent productivity gains.

01

China's Computing Architecture Becomes a Rival

Accelerated computing

IDC data reported by Reuters shows Nvidia's share of China's data-centre accelerator unit shipments fell to 34% in H1 2026, with Huawei at 28%. Nvidia shipped approximately 678,000 units, down 44% year over year; Huawei shipped approximately 572,000, of a total of roughly 2.01 million. Non-GPU accelerators overtook GPUs by volume for the first time.

The shift reflects both US export restrictions and Chinese industrial policy; it measures units, not performance or revenue. China is building commercial scale under restricted access, moving the competitive unit from the individual accelerator toward whole-system performance and economics — hardware, networking, software, inference efficiency and electricity.

The repricing was immediate: Samsung fell approximately 8%, SK Hynix around 10% and South Korea's KOSPI over 6% on October 13, amplified by reports of advances at memory producer CXMT.

02

AI Credit Exposure Turns Systemic

Capital

The GFSR's most consequential finding: hyperscalers' share of investment-grade debt issuance has risen from under 1% in 2024 to approximately 10% of new issuance. The IMF warns adverse AI-profitability developments could trigger asset-price adjustments and financial spillovers.

Tencent began marketing a four-tranche international offering on October 13 that could raise at least $4.5 billion, with 10- and 20-year dollar tranches guided around 115 and 140 basis points over comparable Treasuries. Its first-half capital expenditure reached approximately 96.3 billion yuan, more than double the 2025 level.

The financing market is entering a more selective phase: established, cash-generative issuers retain access to long-duration funding, while less-established developers face materially tighter conditions. The IMF's stress test now finds banks representing roughly 22% of global banking assets with weak capital positions under its adverse scenario, up from 18% a year earlier — a hypothetical result, not a current breach.

03

The Energy Shock Could Delay the Dividend

Energy

The IMF projects global inflation of 4.7% in 2026 and 4.1% in 2027, assuming oil near $96 and $87 per barrel. Under an adverse scenario with oil toward $130 through the first quarter of 2027, growth would fall by approximately 0.5 percentage point and inflation rise by 1.1 points in 2027. Higher energy costs and discount rates together reduce the present value of AI infrastructure cash flows — before the anticipated productivity benefits are realized. The October 6 Google–Constellation agreement (890 MW of incremental nuclear capacity, plus a separate 2,700 MW supply arrangement) remains the structural baseline.

04

The Split Becomes Geographical

Macro

The IMF reports 32 countries now account for 95% of global growth, down from 54 in April. Global trade grew 8.3% between 2023 and 2025; AI-related trade grew 29% — without it, overall growth would have been about 4.6%. The Eurozone forecast is upgraded to 1.1% for 2026, from 0.9%, but remains far from US and Chinese scale. Europe's task is to capture more domestic value added rather than compete only to attract infrastructure commitments. The revised Irish Presidency budget — €1.62 trillion for 2028–2034, 8% below the Commission proposal but roughly 30% above the current framework, with more than €400 billion for competitiveness — heads to the European Council on October 15–16.

Key takeaways

For investors and boards

  • AI has become a macroeconomic growth variable — the improved IMF outlook confirms its weight in aggregate activity.
  • AI financing has become a financial-stability variable — hyperscaler issuance demands rigorous assessment of leverage and concentrated exposure.
  • China's accelerated-computing ecosystem is a material competitive force; distinguish industry growth from individual margin sustainability.
  • Europe must make productivity — not infrastructure announcements — the principal measure of AI competitiveness.
  • The next market rotation could reward productive AI adoption over indiscriminate infrastructure expansion.

The decisive question is whether productivity can grow fast enough to justify the capital structure financing the AI revolution. AI is becoming too important to global growth to remain a purely technological investment — and too capital-intensive to escape the discipline of financial returns.

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