Chinese tech sector’s open-source approach significant for future of AI development, governance: former UN official

Jovan Kurbalija argues that open-source AI — highlighted by Chinese platforms such as DeepSeek (launched Jan 20) — is reshaping a landscape formerly dominated by proprietary models and could realign competitive dynamics across the tech sector. He emphasizes inclusivity (noting one‑third of humanity remains unconnected), the need for education and capacity building, and the role of governance initiatives (e.g., China’s Global Governance Initiative in Sept 2025) in anchoring AI development into societal and cultural frameworks; geopolitical tensions and regulatory choices will determine adoption patterns and regulatory risk exposure for investors.

Analysis

Market structure: Open‑source AI leadership (exemplified by DeepSeek) shifts value toward compute, chips, cloud infra and cybersecurity while compressing licensing/pricing power of proprietary model vendors. Expect data‑center GPU demand to rise ~20–40% over 12 months, benefitting NVDA, AMD and cloud (MSFT, AMZN, GOOGL) and increasing capex intensity for hyperscalers; winners also include cybersecurity names (PANW, FTNT) as attack surface grows.

Risk assessment: Key tail risks are abrupt export controls/sanctions (weeks–months) that could cut chip flows to China, and an AI regulatory shock (3–12 months) that raises compliance costs 5–15% for platform providers. Hidden dependencies include fab capacity (TSMC/SMIC) and power/grid constraints for hyperscale buildouts; catalysts to watch in 30–180 days are DeepSeek adoption metrics (repo forks, downloads), Chinese Global Governance announcements, and US export policy moves.

Trade implications: Tactical overweight semiconductors and cloud infra, underweight pure‑play LLM/SaaS names whose moats are threatened. Implement 6–12 month trades exposed to compute scarcity (long NVDA, AMD; long MSFT/AMZN/GOOGL; long PANW/FTNT) and pair with shorts of high‑multiple AI pure‑plays (C3.ai AI). Use options to express directionally with defined risk (call spreads on NVDA, put protection on AI names).

Contrarian view: Consensus focuses on geopolitics and may underprice the deflationary effect of commoditized open‑source models on software margins; historical parallel is Linux/web protocols where value migrated to services and infrastructure. If open‑source adoption accelerates, expect software multiples to compress 10–30% over 18–36 months while infra and hardware capture the upside.

Key Decisions for Investors

• Establish a 2–3% long position in NVIDIA (NVDA) within 2–8 weeks funded by reducing 1–1.5% exposure to high‑multiple AI pure‑plays; target horizon 6–12 months, add if quarterly data‑center revenue growth >20% or GPU utilization >80%.
• Add 1–2% long exposure to cloud infra (split MSFT, AMZN, GOOGL) and 0.5–1% to cybersecurity (PANW, FTNT); hold 6–18 months and increase allocation if hyperscaler capex guides up by >10% YoY.
• Establish a 1% short position in C3.ai (AI) or equivalent AI‑SaaS names and buy 3–6 month puts as hedge; reduce or close if consensus EBITDA revision improves by >15% or open‑source adoption metrics stall for 60+ days.
• Buy a defined‑risk 9–12 month NVDA call spread (30–50% OTM) sized to 0.5–1.0% of portfolio premium to capture upside from compute tightness; simultaneously buy 3–6 month put protection on AI pure‑plays sized to 0.3–0.6% portfolio premium.
• Monitor (daily→weekly) DeepSeek/GitHub forks, Chinese cloud adoption stats and US Commerce export notices over next 30–90 days; if export controls are announced, trim China‑exposed semiconductor longs by 40% within 48 hours and rotate into domestic AI infrastructure names.