Press Enter to search · ESC to close

Macro

UBS Lifts AI Capex Forecast to $998B for 2026 as Memory Prices Rewrite the Cycle

UBS has raised its 2026 global AI capex forecast to $998 billion, nearly double 2025 levels, with about 90% of the incremental spending through 2027 coming from rising memory costs. The shift implies limited US real GDP impact and a profit transfer to Asian memory producers.

Memory Costs Drive Nearly All of the Incremental Spending

UBS has sharply raised its forecast for global AI capital expenditure, projecting that total spending will reach $998 billion in 2026, nearly double the $506 billion expected in 2025. By 2027, the figure is forecast to climb further to $1.447 trillion. The revision is driven overwhelmingly by rapidly rising memory prices rather than a broad expansion in infrastructure investment volumes.

The distinction matters. UBS estimates that global memory spending will jump from $71 billion in 2025 to $367 billion in 2026, then to $923 billion in 2027. Over the same period, non-memory AI capex is expected to peak at roughly $631 billion in 2026 before declining to $525 billion in 2027. Memory’s share of total AI capex is therefore set to rise from about 14% in 2025 to 37% in 2026 and 64% in 2027.

Where the Incremental Dollars Come From

UBS’s decomposition shows that memory cost inflation accounts for roughly 60% of the year-over-year increase in AI capex in 2026. In 2027, memory spending growth actually exceeds the net increase in total AI capex, because spending on other components falls over the same period. Taken together, UBS calculates that about 90% of the nearly $1 trillion in incremental AI capex between 2025 and 2027 will come from higher memory spending.

This is a structural shift in the economic logic of the AI investment cycle — from pure scale expansion of infrastructure toward the rising cost of critical components needed to support increasingly powerful computing systems.

Market Implications

Equities

  • Asian memory producers are the clearest beneficiaries. If the spending increase is price-driven, the money flows as a profit transfer from AI infrastructure investors to memory manufacturers, most of which are based in Asia. Korean and Japanese memory names, along with their suppliers, stand to capture a disproportionate share of the value.
  • US AI infrastructure and hyperscaler names face a more nuanced picture. Higher component costs pressure margins and could slow the pace of capacity additions, even as headline capex numbers rise.
  • Non-memory component suppliers may see order volumes decline into 2027 as budgets are reallocated toward memory.

Macro and GDP

UBS explicitly notes that the nature of memory spending — price versus volume — has fundamentally different macroeconomic consequences. If incremental spending mainly reflects higher prices, the contribution to US real GDP will be limited. Instead, the flows represent a transfer of income and profits to Asian memory producers, providing a more direct GDP boost to those economies. This suggests the AI boom’s measured US growth impact may be smaller than headline capex figures imply, while Asia’s export and industrial output data could surprise to the upside.

Currencies and Rates

  • Asian currencies — particularly the Korean won and Taiwanese dollar — could find support from stronger export earnings and terms-of-trade gains.
  • The US dollar may see less support from AI-driven investment flows than previously assumed, though the net effect depends on how much of the spending is ultimately financed domestically.
  • Bond markets could read slower real GDP impulse as mildly disinflationary for the US, though the capex headline remains supportive of risk appetite.

Commodities and Crypto

  • Semiconductor materials and specialty gases tied to memory fabrication should see firm demand.
  • Crypto markets may take a mixed signal: continued AI capex enthusiasm supports the broader risk-on narrative and AI-adjacent token narratives, but a rotation of value toward Asian hardware producers could divert speculative flows away from digital assets.

Key Takeaways for Investors

  • The AI capex story is no longer just about building more data centers — it is increasingly about the cost of memory. Portfolios positioned for volume growth may need to adjust toward pricing power.
  • Asian memory supply chains deserve a closer look. UBS’s framework implies their earnings and GDP contribution could exceed consensus expectations.
  • US real GDP may get less of a boost than the headline capex numbers suggest, which has implications for rate expectations and dollar strength.
  • Watch the price-versus-volume split closely. If memory prices normalize, the entire capex forecast could be revised down sharply — a key risk to the current narrative.

View original

Share
Risk notice This site provides news and information on the crypto, blockchain and Web3 industry for reference only and does not constitute investment advice or any promise of returns. Virtual currency-related activities are illegal financial activities in mainland China; digital asset prices are highly volatile; use at your own risk. This site does not provide trading, token issuance or related referral services.

Related Reading

Latest News

TREE NEWS share card
Long-press image above → Save to Photos / Share
Pitch us Feedback