Benchmark’s Bill Gurley: the AI bubble is about to burst—and a reset is coming

Bill Gurley warns of an impending AI bubble after the AI boom made the world’s 500 wealthiest people $2.2 trillion richer in 2025, saying excessive AI-driven spending and valuation inflation will force a reset. Morgan Stanley projects capex-to-sales rising to 34% this year and 37% by 2028 (potentially 38%/45% with hyperscalers), with ~$2 trillion of spending from 2026–2028 and nearly $1 trillion in undisclosed future lease commitments (about $662B not yet commenced). High-profile SaaS names (Salesforce, ServiceNow) have lost >20% since early 2026, companies are cutting headcount and executives may blame AI for over-bloated operations, while OpenAI/Anthropic funding and cash-burn forecasts (OpenAI ~$207B additional funding need to 2030; ~$280B estimated cash burn) highlight sustainability risks.

Analysis

The immediate market dynamic will be a liquidity-driven stress test rather than a pure technology verdict: firms that funded growth with long-dated contractual obligations or aggressive sales spend will show volatility first as funding and credit tighten. That makes high CAC, multi-year revenue-recognition SaaS models particularly fragile because unit economics reprice slowly while cost of capital reprices quickly, compressing LTV/CAC and forcing margin remediation.

Hidden lease-like commitments and vendor financing create quasi-leverage on the largest cloud providers and their partners; those off-balance obligations effectively shorten the runway of startups and raise counterparty risk for suppliers of racks, networking gear and specialised fabs. When utilization or pricing power falters, order books and regional construction activity will flip from boom to bust within a single reporting cycle, producing outsized EPS downside for exposed suppliers.

Key catalysts to watch on a 3–24 month horizon: tightening in venture rounds and secondary financing terms, tranche-realization dates for undisclosed lease commencements, and sequential guidance downgrades from enterprise software vendors signaling durable margin compression. Reversal of the current narrative requires demonstrable, margin-accretive monetization of AI services (per-seat or vertical SaaS with sticky upsells) or a sustained fall in funding costs that restores long-duration valuations.

The actionable framing is a two-step trade: front-run a liquidity/valuation reset with asymmetric downside protection, then scale into high-quality SaaS on clear remediation signals. Expect a rapid rotation opportunity lasting weeks after the first meaningful guidance cuts—this is when crowded short positions and option flows create convex rebounds for select survivors.

Key Decisions for Investors

• Short CRM & NOW via 3–6 month put spreads (buy 3-month 25–30 delta puts, sell nearer OTM puts to finance) sized 1–2% NAV each. Rationale: immediate re-rate risk from fixed-cost sales models; target 30–45% downside in 3–6 months; loss limited to net premium paid.
• Initiate a 12–18 month protection trade on AMZN + MSFT (60/40) using bought put verticals (dec-2027 20–30% OTM put spreads). This is a hedge against lease/hidden-liability-driven deleveraging across hyperscalers; aim for 20–30% move in 12–18 months, with defined max loss = premium paid.
• Set conditional buy orders to accumulate CRM & NOW (or 12–24 month LEAPS calls) only upon: (a) two consecutive quarters of negative organic revenue guidance OR (b) an additional 35–45% price decline. Size initial entry 2–3% NAV each and scale to 6–8% if remediation signals (margin cuts + S&M rationalization) occur—target asymmetric 3–5x upside post-reset.
• Tactical long MSFT via 12–18 month call spreads as a defensive overweight (buy dec-2027 10–15% ITM calls, finance with nearer-term call sales). Rationale: diversified enterprise cashflows and balance-sheet optionality make it a lower-risk way to capture AI monetization upside while limiting premium outlay; expect lower volatility drawdown vs pure-play hyperscalers.