'Kingdom Come: Deliverance 2' translator says developers fired him to use AI

Warhorse Studios notified longtime English translator Max Hejtmánek on March 27, 2026 that his role will be made 'obsolete' and replaced by AI for all translations effective next month; Hejtmánek, who worked at the studio since 2022, announced the layoff on Reddit. The decision likely yields short-term cost savings for the studio but creates reputational and product-quality risks for the game and highlights broader labor-disruption risk across creative industries as AI adoption accelerates.

Analysis

AI-driven localization will lower explicit labor costs but creates an outsized risk to narrative integrity and user trust that is measurable inside the product funnel: expect degraded review scores and reduced playtime in story-heavy titles to show up within 1–2 release cycles (6–18 months), translating into a 3–8% hit to title lifetime revenue if unaddressed. The real second-order victims are specialist supply-chain nodes — boutique localization houses, voice-over agencies, and localization QA teams — whose volume is addressable and thus vulnerable to consolidation or margin compression of 30–60% over 12–24 months.

Winners will be platform and infrastructure firms that sell the AI stack and monitoring tools: recurring-revenue LLM API providers, GPU vendors, and telemetry/QA analytics vendors that can quantify linguistic degradation and flag hallucinations. A parallel arbitrage is emerging: buyers of low-cost AI localization may realize short-term SG&A savings (~0.5–2% of dev budgets) but suffer longer-term engagement losses that exceed those savings, creating acquisition targets for firms offering “human-in-the-loop” guarantees.

Key catalysts and risks: fast adopters will report immediate cost savings (near-term positive headlines), but the reversal triggers are consumer backlash (spikes in negative reviews within weeks of release), regulatory/rights litigation over training data in 6–24 months, or high-profile franchise miss that forces studios to rehire experts. Tail risk: a major AAA franchise materially underperforms due to poor localization, prompting industry-wide reversion to humans and a sharp premium revaluation for specialist providers.

Contrarian angle: the market’s binary framing — human obsolete vs fully automated — is overstated. Expect a durable premium for validated human review and creative editing (2–4x pricing power vs raw AI output). That creates a profitable bifurcation: high-margin verification and QA tools on one side, commoditized bulk translation on the other.

Key Decisions for Investors

• Long NVDA (6–18 months): buy calls or add exposure to capture continued GPU-driven AI demand. Rationale: infrastructure is a direct beneficiary of enterprise LLM adoption; reward >2x if adoption accelerates, downside limited to semiconductor cyclicality (30–40% drawdown risk).
• Long MSFT or GOOGL (12 months): accumulate shares or buy 9–12 month call spreads to play enterprise LLM and API monetization. Risk/reward: steady recurring revenue lift if studios adopt managed LLM products; downside is slower AI uptake or margin pressure from competitors (~15–25% downside).
• Short select mid/small-cap narrative game developers (3–12 months): pair short against a large-cap studio (pair: short CDR.WA or similar narrative-heavy mid-cap / long ATVI or TTWO). Rationale: smaller studios rely disproportionately on quality localization; poor execution can knock 3–8% off title revenues. Tail risk: successful PR-managed turnarounds or buyouts.
• Long LSE:RWS or APX (Appen) equivalents (6–18 months): accumulate exposure to firms offering human-in-the-loop localization and QA. Rationale: these firms can reprice to a premium as customers demand verified quality; target upside 30–60% if industry rebalances, risk is further automation and margin compression (20–35% downside).