Most important take away

Large US tech companies are showing classic price inelasticity in AI investment — they keep accelerating spending on chips, power, and data centers despite massive cost increases in inputs (copper +40%, gas turbines +50%, memory +150–300% YoY). This is buttressing US growth and equity earnings, but also risks fueling inflation and pushing corporate bond spreads wider as record debt issuance continues.

Summary

Andrew Sheets (Morgan Stanley, Global Head of Fixed Income Research) frames the AI buildout as a study in price inelasticity. Demand for AI infrastructure is holding up — and accelerating — even as the prices of the underlying components surge. Morgan Stanley estimates roughly $800B of AI-related capex from large US tech companies this year (about 2x 2025 and 3x 2024), with forecasts continuing to be revised upward toward $1.1T in 2027.

Actionable insights and investment implications:

• Equity exposure: This inelastic spending supports US growth despite slowing job creation and ongoing geopolitical uncertainty (Iran). It also underpins the bullish 2026 earnings call from Morgan Stanley’s US equity strategist Mike Wilson — implying continued support for large-cap tech and AI-levered names.
• Commodities / suppliers: Suppliers of inelastically demanded AI inputs are direct beneficiaries — copper (up ~40%), gas turbines (up ~50%), and memory chips (up 150–300% YoY). Names exposed to these supply chains have pricing power.
• Power / data center infrastructure: Gas turbine makers and electrical/power infrastructure providers are seeing strong pricing — a tailwind for the power-equipment and utility-adjacent complex.
• Credit markets — caution: Tech hyperscalers are issuing debt at a record pace even as their borrowing costs rise. Sheets and Morgan Stanley continue to expect record supply and modest spread widening in US investment-grade corporate bonds. The implication: be cautious on IG credit exposure, particularly tech-heavy issuance, as sympathetic widening could spill to other issuers.
• Inflation risk: Persistent demand at rising prices feeds into inflation while core inflation already sits above the Fed’s target — a risk for rate-sensitive assets and a reason to be wary of duration.
• Return uncertainty: It remains unclear what returns this historic capex will generate. Investors should weigh the durability of AI monetization against the sheer scale of spend when sizing AI-related positions.

No specific tickers were mentioned, but the discussion implicates the large US tech hyperscalers (the megacap AI capex spenders), memory suppliers, copper producers, gas turbine manufacturers, and the broad US IG corporate bond market.

Chapter Summaries

• Intro and concept of elasticity: Sheets opens with the economics 101 idea of price elasticity — pizza vs. electricity or concert tickets — to introduce inelastic demand.
• AI as the inelastic case: AI capex is central to current market conversation, supporting US growth and stock earnings despite macro headwinds.
• Scale of the buildout: ~$800B in 2026 from large US tech (2x 2025, 3x 2024), rising to an estimated $1.1T by 2027.
• Component prices surging: Copper +40%, gas turbines +50%, memory +150–300% — yet demand keeps accelerating.
• Why the price insensitivity: Companies view AI as the most important technology in a decade, have ample financial resources, and are willing to pay up to compete for an uncertain winner.
• Implications — positives: Real commitment to AI supports earnings (aligning with Mike Wilson’s US equity view) and overall growth.
• Implications — risks: Inflationary pressure on already-elevated core inflation, record debt issuance widening corporate bond spreads, and uncertain returns on the historic investment.
• Closing: Morgan Stanley maintains a call for record supply and modest spread widening in US IG corporates.