The New Map of AI Power

Summary

Morgan Stanley's Ariana Salvatore and Stephen Byrd discuss AI sovereignty, US and China approaches, and the investment implications of a more fragmented AI ecosystem. They argue sovereign AI reinforces the AI capex cycle by making infrastructure more redundant and capital-intensive. Key beneficiaries include semiconductors, data centers, networking, power, cloud, cybersecurity, and infrastructure software, with specific advantages for collocation operators and behind-the-meter or off-grid power. China's push for a lower-cost indigenous AI ecosystem is flagged as a structural shift to monitor.

• AI sovereignty involves government control over compute, data, energy, and technology amid rising geopolitical tensions.
• The US is pursuing a middle path between security guardrails and global AI reach, while China is building an indigenous AI ecosystem.
• Fragmentation could make AI infrastructure more duplicated, capital-intensive, and potentially inflationary.
• Named beneficiaries include semiconductors, data centers, networking, power, cloud, cybersecurity, and infrastructure software.
• Localization and data-sovereignty needs specifically favor collocation operators.
• Energy constraints and grid pressures increase interest in behind-the-meter and off-grid power solutions.
• Sovereign AI is framed as reinforcing the multiyear AI capex cycle rather than undermining it.