New report claims Kingdom Come: Deliverance 2 studio fired translator to replace them with AI and "save finances"

A translator at Warhorse Studios was reportedly fired on 27 March after roughly 3.5 years and told their role was 'obsolete' as the studio plans to replace translation work with AI to 'save finances', creating reputational and employee-relations risk. While isolated, the case highlights a wider industry pivot to AI (citing a recent Pearl Abyss disclosure) that could increase regulatory, PR, and talent-retention pressures for game developers.

Analysis

The immediate economic vector is predictable: generative models compress unit cost of localization work quickly while raising marginal costs for quality assurance and IP compliance. Expect studios to realize step-function labor savings in 3–9 months for iterative assets (patch notes, UI strings) and broader displacement across full-game localization over 12–36 months once tooling, pipelines, and post-editing workflows are standardized.

Second-order winners include GPU and cloud compute providers that sell the training/inference stack, and middleware AI vendors that sell orchestration/guardrails (where recurring SaaS margins scale). Losers are legacy localization integrators and bench-heavy QA shops whose revenue and margins can be compressed 20–40% unless they pivot to higher-value human-in-the-loop certification or regulatory-compliance services.

Operational and governance risks create asymmetric payoff scenarios: a high-profile mistranslation or IP/data-leak tied to model training can generate reputational damage, regulatory scrutiny, or class-action exposures that hit revenue and force costly remediation (manual fixes, re-localization). The most likely catalysts to reverse or slow the move are consumer backlash (platform review impacts within days-weeks), unionization/collective bargaining outcomes (months), or rapid regulation around AI training data/privacy (12–24 months).

Key Decisions for Investors

• NVDA (long, 6–12 months): overweight NVDA to capture sustained GPU demand for on-prem and cloud model inference. Position size: tactical overweight ~3–5% of liquid equity sleeve; target 25–60% upside if AI capex continues, stop -20% on thesis failure (AI spending pause).
• GOOGL (long, 6–12 months): buy GOOGL to play enterprise and consumer translation monetization and model deployment; consider a 9–12 month call spread to define downside. Risk/reward: asymmetric — limited premium loss vs ~20–40% upside if adoption accelerates and ad/Cloud monetization follows.
• TIXT / KWS.L (short, 6–18 months): initiate selective short exposure to pure-play localization/QA vendors (TELUS International TIXT or Keywords Studios KWS.L) where >20% of revenue is human localization. Thesis: 20–40% revenue compression risk over 12–24 months; use tight stops (+15%) and size as a hedge to long AI infra exposure.
• Pair trade (6–12 months): long NVDA + GOOGL vs short TIXT (notional 2:1 on longs:short): captures infrastructure upside while hedging industry-specific disintermediation risk. Rebalance monthly; trim longs if headlines reveal major legal/regulatory rulings or if price action outperforms fundamentals by >30%.