From this lens, let’s explore the implications,
The DeepSeek Problem is No-More
Technological nationalism, specifically the idea that AI software like DeepSeek (a Chinese-developed model) might be restricted or banned from U.S. use.
???? AI, Software, and Data as Strategic Assets
In this worldview, AI isn’t just a tech tool — it’s a national security asset.
So, when we say tools like DeepSeek, ChatGLM, or any LLM trained in China may not be allowed in the U.S., that fits a broader strategic decoupling in software:
???? Why DeepSeek and similar software might be restricted:
- Data origin: U.S. institutions can’t trust models trained on data from unknown or adversarial sources.
- Alignment risks: LLMs reflect the values of their training set and system design — could be misaligned with U.S. security, cultural, or legal norms.
- Industrial espionage & IP leakage: Risk of software transmitting or learning from sensitive or proprietary data.
- Military crossover: Dual-use concerns — LLMs can be used in cyber, drone, or psychological operations.
???? How This Changes the Investment Landscape
- Domestic AI Firms (Winners)
- Palantir (PLTR) – Government-aligned AI and defense software
- Nvidia (NVDA) – Hardware foundation for domestic LLMs
- Databricks (if public), Snowflake (SNOW) – Domestic data infra
- AI Sovereignty Themes
- Rise of U.S.-only LLMs (Anthropic, OpenAI, Mistral US-hosted)
- Government contracts require origin transparency
- Enterprise AI adoption tied to geopolitical “trust scores”
- Likely Regulation Path
- Full auditability of AI models used in U.S. infrastructure
- Origin labelling (like food or textiles) for software
- Prohibition or license-control for foreign LLMs (especially from China or adversarial states)
Discover more from Priory House
Subscribe to get the latest posts sent to your email.