KKR says AI productivity boom to keep on going — but warns of 'extreme' trend not seen since the 19th century

KKR says the AI-driven productivity boom is still early, but growth may become more concentrated across a few industries, with technology, high-end services, government spending, defense, and power identified as relative winners. The firm also sees Japan and Korea as cheap and expects the Chinese yuan to strengthen toward about 6.5 per dollar by 2027 as the U.S. dollar peaks. It highlighted wheat as a strategic, policy-backed commodity, citing USDA forecasts for 2026-2027 U.S. production at the lowest since 1972 and prices at three-year highs.

Analysis

The key implication is not simply that AI lifts productivity, but that it widens dispersion in capital allocation and margin power. That favors businesses with scarce inputs—power, chips, defense-adjacent capacity, and regulated infrastructure—while compressing returns in labor-heavy, easy-to-replicate sectors where AI becomes a margin tax rather than a growth driver. The second-order effect is a winner-take-most dynamic in public markets: index-level earnings may hold up, but breadth likely stays poor, which is constructive for dispersion trades and selective long/short book construction.

The macro setup also argues for a prolonged capex cycle in electricity, grid equipment, and resilience assets. AI compute demand is already forcing a re-rating of power availability as a strategic bottleneck, which should support utilities with rate-base growth, independent power producers with flexible generation, and industrials tied to transformers, switchgear, and backup systems. Defense remains structurally bid because the same fragmentation driving supply-chain reshoring also increases sovereign demand for domestic capacity, munitions, and cyber/space infrastructure.

On Asia, the more interesting read is that the earnings surprise potential in Japan and Korea is less about cyclical recovery and more about operating leverage from governance reform, weak historical expectations, and AI-linked hardware exposure. That makes the region a cleaner beneficiary than China, where property deleveraging still acts as a valuation anchor and caps the transmission of global AI capex into broad domestic demand. FX is the underappreciated lever: a stronger yuan should help translate local asset performance into dollar returns, but only if the dollar’s peak is durable rather than a short pause.

Key Decisions for Investors

• Add a tactical long basket of power infrastructure beneficiaries (ETN, HUBB, PWR) vs short labor-intensive industrials (fast-cycle services/contractors) over the next 3-6 months; thesis is that AI-related grid capex and electrification keep order books firm while margin pressure rises elsewhere.
• Buy defense exposure on pullbacks via LMT/RTX or XAR for a 6-12 month hold; risk/reward improves if geopolitical fragmentation persists and procurement budgets stay insulated from broader growth softness.
• Initiate a relative-value long Japan/Korea vs short China equity exposure using EWJ/KOR vs FXI or KWEB over 6-12 months; this captures upside from earnings revisions and governance re-rating while avoiding the property overhang.
• Consider long CNH vs USD via FXI? Actually use a pure FX proxy or CNH-linked exposure if available; timeframe 6-18 months, with upside if dollar peaking thesis holds, but size modestly because the catalyst is macro and can be delayed.
• Express the dispersion view with a long QQQ / short equal-weight or cyclical broad-market hedge only if breadth deteriorates materially; near-term reward is strongest if mega-cap AI leadership persists while the rest of the market lags.