Goldman investment banking co-head Kim Posnett on the year ahead, from an IPO ‘mega-cycle’ to another big year for M&A to AI’s ‘horizontal disruption’

Goldman Sachs' Kim Posnett says AI has moved from experimentation to industrialization and is the primary catalyst for a surge in strategic dealmaking and capital markets activity in 2026. Global M&A reached $5.1 trillion in 2025 (up 44% YoY) and sponsor dynamics remain powerful with ~$1 trillion of dry powder and over $4 trillion of unmonetized portfolio companies, underpinning expectations for a large IPO 'mega-cycle' driven by institutionally mature, capital-intensive private tech giants; firms must also navigate divergent US, EU and UK regulatory regimes when deciding where to build, deploy and transact.

Analysis

Market structure: AI industrialization disproportionately benefits cloud and enterprise-stack owners and advisors — think GOOGL and ORCL for infrastructure and GS for advisory fees — as corporates accelerate capex for models and agents. Winners also include software companies that convert AI into recurring SaaS value; losers are hardware-centric consumer plays and regionally constrained EU incumbents facing guardrails that compress addressable markets. Cross-asset: equity risk-on should tighten corporate spreads and compress 2–5yr Treasury yields by 25–75bp on rotation into growth; USD directionally stronger near-term if the Fed resists easing, but divergent policy (US pro-growth vs EU guardrails) favors USD and US-listed tech FX flows.

Risk assessment: Tail risks include abrupt EU/UK enforcement that fragments data flows, a major model safety incident leading to heavy regulation, or compute chokepoints (GPU/power) that push up costs 30–100% for large trainers. Immediate (days) risks: deal/IPOs volatility and headline-driven repricings; short-term (3–6 months): M&A waves and sponsor exits; long-term (12–36 months): structural winner-take-most consolidation. Hidden dependencies are physical compute (datacenter power/semis), talent wage inflation, and geopolitical export controls that can suddenly raise operating costs or delay deployments.

Trade implications: Direct plays — establish tactical 2–3% long in GS (trade execution/advisory exposure) and 3–4% long in GOOGL for AI infrastructure exposure, scaled over 4–8 weeks. Pair trade — long ORCL (2%) vs short AAPL (1–2%) for 6–12 months: ORCL captures enterprise AI lift, AAPL at risk of slower monetization; use 3–6 month call spreads on GOOGL to cap premium and buy protective puts on positions if index drops >10%. Rotate portfolio +10–15% overweight into Cloud/Enterprise Software, underweight Consumer Hardware for H1 2026; add on M&A or IPO pullbacks over the next 3 months.

Contrarian angles: Consensus overlooks compute and power as rate-limiting factors — if GPU supply tightens or power costs rise 20–40%, ROI timelines stretch 12–36 months and valuations re-rate. The IPO “mega-cycle” may be front-loaded: large listings could depress post-IPO returns as index flows dilute near-term; conversely, open-source models (DeepSeek-style) could compress moats quicker than priced in, benefiting infrastructure but hurting proprietary model vendors. Expect talent-driven margin pressure and regulatory segmentation that creates durable dispersion between US winners and EU/UK laggards.

Key Decisions for Investors

• Establish a staggered 2–3% long position in GS over 4 weeks to capture advisory fee tailwinds from IPOs/M&A; trim if GS rallies >15% from entry or if M&A headlines disappoint over 60 days.
• Build a 3–4% core long in GOOGL (scale into weakness) with a 6–12 month horizon; hedge with 3–6 month put protection if position falls >12% to protect against IPO-season volatility.
• Initiate a 2% long ORCL / 1.5% short AAPL pair (net 0.5% long) for 6–12 months to play enterprise AI adoption vs consumer hardware monetization; close or rebalance if ORCL outperforms AAPL by >20% or underperforms by >10%.
• Purchase 3–6 month call spreads on GOOGL (buy ATM, sell +15% strike) sized to 0.5–1% portfolio risk to participate in near-term AI announcements while capping premium outlay.
• Reduce sector exposure to Consumer Hardware/Devices by 5–10% and reallocate into Cloud/Enterprise Software and AI-infrastructure suppliers over the next 1–3 months; revisit after major IPOs or if GPU availability improves by >25%.