Former Baidu President on AI Tokenization in China

Zhang Yaqin, former Baidu president and head of Tsinghua's Institute of AI Industry Research, said AI tokenization is "exploding" in China and has outpaced examples like OpenClaw earlier this year. This suggests accelerating tokenization and digital-asset activity that could benefit AI infrastructure, cloud providers, fintech and crypto-service firms in China. Near-term market impact is limited absent quantification or regulatory clarity, but investors should monitor adoption and policy developments closely.

Analysis

Tokenization of AI outputs is evolving into an infrastructure-led growth vector rather than just a product gimmick: if developers and enterprises tokenize model outputs or access rights, cloud-hosting and secure runtime (TEE/SGX-style) fees become recurring, high-margin revenue rather than one-off API calls. That creates a 12–24 month runway for outsized capex cycles in data centers and AI chip supply — expect 15–30% incremental capacity demand in targeted AI clusters before broad consumer adoption arrives.

Winners are the stack providers who control runtimes, custody and settlement — domestic AI-chip vendors, cloud/IDC operators, and regulated digital-asset custodians — because they capture the fixed-cost leverage of token-enabled monetization. Incumbent consumer platforms face a two-way dynamic: they can re-issue tokens and keep economics, or lose ad/transaction monetization to decentralized marketplaces; the latter would compress CPM-style monetization by mid-single digits to low-double digits over 2–3 years for exposed ad businesses.

Key risks are regulatory and supply-side: an authoritative PBOC-like clarification, provincial sandbox pullback, or tightened export controls on advanced GPUs can unwind enthusiasm in weeks to months. Conversely, province-level pilot approvals, partnership announcements with state-owned financial infra, or a marquee token issuance by a major enterprise are 3–9 month catalysts. Contrarian view: consensus bullishness underprices friction — tokenization is more likely to see concentrated B2B adoption with high compliance overheads than rapid consumer ubiquity, meaning infrastructure winners will be fewer and more capital-intensive than the market currently assumes.

Key Decisions for Investors

• Buy a 9–12 month BIDU call spread (size ~1% NAV) to capture optionality from cloud/AI monetization while capping premium outlay; target ~3x upside if Baidu captures token issuance flows across search/AI services, downside limited to premium and regulatory shock risk — monitor on any provincial pilot announcements.
• Buy NVDA 6–12 month calls (size 0.75–1.5% NAV) to express incremental AI compute demand; hedge ~20% of notional by shorting Tencent (TCEHY) 6–9 month calls to protect vs China-specific regulatory/ad-revenue stress that could temporarily dent compute demand — exit or re-hedge on clear US export-control developments.
• Long Super Micro (SMCI) or other AI-focused server hardware exposure (6–18 month horizon) via stock or call position (size 1% NAV) to ride data-center hardware re-rates; size conservatively given potential bottlenecks in advanced node chips and logistics.
• Initiate a small, tactical short of ad-dependent China platforms (eg. TCEHY/BABA, total size 1% NAV) financed by the BIDU call spread premium — thesis: tokenization/marketplace shifts will compress ad monetization over 12–24 months unless platforms capture issuance rights; tighten if regulatory guidance favors platform-controlled token issuance.