Owlcat Games is using generative AI to 'iterate faster' on The Expanse

Owlcat Games confirmed it is using generative AI for prototyping and placeholder assets on its upcoming RPG The Expanse: Osiris Reborn and stated all AI-generated placeholders will be replaced and final in-game assets will be 100% human-made. This disclosure is primarily reputational/operational—unlikely to move financials or stock prices materially—but it underscores broader industry adoption of genAI and modest disclosure risk following similar incidents at Pearl Abyss and Capcom.

Analysis

GenAI as a rapid prototyping layer materially compresses early-stage creative cycles: a conservative estimate is 20-40% faster concept iteration, which on a 12–18 month project can shave 2–6 months from preproduction and concentrate spend into later, higher-value phases (QA, narrative polish). That timing shift favors studios that can rapidly iterate and test player-facing changes, increasing the value of real-time engines, asset-pipeline automation, and cloud inference capacity.

Leakage of prototype assets into released builds is the primary operational risk and has real P&L consequences. Even a single high-profile leak can force emergency rollbacks, refunds, and legal defenses; model training/data-use litigation or takedowns could produce mid-single-digit to low-double-digit million-dollar hits for mid-sized studios within 0–12 months and create longer reputational drag.

The clear beneficiaries are tooling and infra providers that sit between artists and engines: low-latency GPUs, cloud inference, creative SaaS, and integrated middleware that governs provenance and auditing. Conversely, manual art-outsourcing firms and boutiques that cannot integrate genAI will face margin compression; studios that credibly certify 100% human-made final assets may be able to monetize trust as a marketing premium in the next 6–18 months.

Regulatory and labor responses are the wildcards over 12–36 months: artist unions, IP-rights enforcement, or disclosure rules could raise compliance costs and slow adoption. A contrarian point: the market underestimates monetizable trust — a small cohort of studios promising fully human-made finals could command pricing/PR advantages, creating a bifurcated ecosystem rather than uniform displacement of artists.

Key Decisions for Investors

• Long NVDA (NVIDIA) 6–9 month calls: buy ATM or modest OTM calls to capture incremental GPU/cloud demand from studios adopting genAI pipelines. Target 30–50% upside if adoption accelerates; cut losses at a 40% premium decline. Rationale: direct exposure to inference hardware and cloud GPU demand.
• Long ADBE (Adobe) 9–12 month calls or shares: Adobe's Creative Cloud and Substance lines are positioned to monetize studio-level genAI workflows. Target 20–30% return over 12 months; stop-loss at 15% drawdown. Rationale: subscription upsell and enterprise tooling for provenance and asset management.
• Long UNITY (U) stock with a 3–6 month protective put (collar): buy U to play embedding of genAI authoring features into engines while hedging governance/monetization risk. Target 25% upside in 6–12 months; collar limits downside to ~10–15%.
• Pair trade: Long NVDA vs Short ATVI (Activision Blizzard) 6–12 month: long NVDA for structural infra gains, short ATVI to express relative PR/regulatory sensitivity among large publishers experimenting with genAI. Aim for >1.5:1 asymmetric payoff; tighten short if EA/ATVI announce credible governance frameworks that repair sentiment.