UBS: Will AI Disrupt Japan's Domestic Software Industry?
Summaries are public. PDF access requires an active membership; all members have the same access. Sign in
UBS examines whether coding AI tools such as Anthropic's Claude Code and OpenAI's Codex pose a structural threat to Japan's IT services industry, which is currently growing at an annual rate of 6–7%. The bank adopts a cautious bearish stance, highlighting three principal risks to the sector.
First, each new coding AI model release brings technological disruption. Frontier models under development using GB200 and GB300 GPUs are expected to be released around October–December 2025, with GPU architectures evolving through Blackwell, Rubin, Rubin Ultra, and Feynman. These advances could render traditional systems integrator development methodologies obsolete, potentially allowing a single person supported by AI agents to complete entire projects.
Second, clients may shift from outsourcing toward internal development as agentic AI lowers barriers. Natural language development reduces engineer requirements, while costs fall well below typical systems integrator monthly fees.
Third, accelerating cloud migration may depress short-term margins due to retraining investments, migration-stage discount pricing, and differing revenue recognition models. Within Fujitsu's Uvance business, vertical offerings exceed 40% gross margins while horizontal offerings fall below 30%.
Bulls counter that AI-driven efficiency gains could expand profitability, with pricing shifting from man-month rates toward consulting-based models. IT budgets may also expand through Jevons' paradox and legacy system renewal needs.
UBS acknowledges that while sustaining strong profit growth appears difficult, evidence for outright profit declines remains insufficient.