‘The Comeback’ reviews: Lisa Kudrow ‘astounds’ in ‘spectacular’ final season warning of AI

Season 3 of HBO's The Comeback premieres March 22 (series finale May 10) and centers on an AI-written sitcom plotline that drives the season's stakes. Critics largely praise the final chapter and Lisa Kudrow's performance as a standout, calling it timely and sharper than prior seasons, though several reviewers flag a darker, more pessimistic tone that reduces comedic payoff for some viewers.

Analysis

The rise of AI-generated entertainment is bifurcating value in the media stack: commodity infrastructure (GPUs, cloud AI) and tooling (transcription, voice‑clone detection, watermarking) will capture the bulk of near-term spend, while legacy IP owners sit on newly monetizable assets that require contract and tech scaffolding to extract value. Expect large studios to reprice a portion of their content libraries as licensing or training-data revenue streams over 12–36 months; even a 1–2% incremental monetization of a $50B aggregate content asset base equals low‑hundreds of millions of annual EBITDA for the largest owners.

Second‑order winners include vendors of provenance and rights management — firms that can embed immutable watermarks, provide chain‑of‑title metadata, or offer auditable consent ledgers — because every studio and distributor will need to reduce litigation and brand risk without slowing distribution. Conversely, pure ad‑platforms and aggregators whose UX depends on user trust (ad quality, brand safety, creator attribution) face a non‑linear revenue hit if AI content amplifies deepfake or low‑quality inventory; ad buyers historically reprice inventory rapidly when viewability or brand safety metrics deteriorate.

Catalysts to watch in the next 6–24 months are: major licensing deals (studios → AI training houses), high‑profile IP litigation outcomes, and any regulatory guidance on synthetic media; each could re‑rate content owners positively or accelerate flight to provenance vendors. The contrarian case is that the market is underestimating downstream legal/clearing costs — if rights monetization proves administratively heavy, incumbents with distribution scale (not just libraries) will win, compressing margins on catalog licensing and favoring vertically integrated players.

Key Decisions for Investors

• Long NVDA (12–24 months): buy equity or LEAP calls to capture sustained GPU demand from media AI training. Rationale: secular GPU capacity tightness; risk: faster competition from custom silicon or cyclical datacenter spend. Target: +30–50% on successful adoption, time to realize 12–24 months.
• Pair trade — Long DIS (6–12 months) / Short ROKU (6–12 months): buy the IP owner, short the ad‑dependent aggregator. Rationale: DIS can monetize legacy IP via licensing or branded AI experiences; ROKU exposed to ad‑quality flight and attribution risk. Risk/reward: asymmetric — downside limited by DIS diversification; short ROKU expects 15–30% downside if ad demand re‑prices.
• Tactical buy VERI (Veritone) or DLB (Dolby) (6–18 months), small position size: these firms provide verification/watermarking and metadata tools. Rationale: rising demand for provenance reduces litigation risk. Risk: execution and scale; size small, target 2–3x upside if product adoption accelerates.
• Options spread on WBD (9–12 months): buy a call spread to express licensing upside while capping cost, funded by selling nearer‑term calls. Rationale: allows participation if IP licensing deals materialize. Risk: headline volatility and slower-than-expected monetization; structure to limit max loss to premium paid.