Who could win and lose on tighter Trump AI controls
CIO Daily Updates
CIO Daily Updates
From the studio
Podcast: Jump Start: Tariff deadline, the Big Beautiful Bill, and US labor resilience (4:57)
Video: Chief economist Paul Donovan: OBBBA winners and losers (12:06)
Video: CIO mid-year checklist– key portfolio steps with Jon Gordon (5:50)
Video: CIO mid-year checklist – equity moves with Crystal Zhao (4:41)
Video: CIO mid-year checklist – credit strategies with Timothy Tay (3:04)
Thought of the day
The Trump administration is preparing new restrictions on AI chip exports to Malaysia and Thailand, according to Bloomberg, with the aim of preventing advanced processors from reaching China via third countries. The draft Commerce Department rule would target shipments from companies like NVIDIA, following concerns over semiconductor smuggling, as well as Chinese access to AI compute capacity based in Southeast Asia. Malaysia has quietly emerged as a key AI data center hub, accounting for nearly 12% of NVIDIA's global sales last quarter. NVIDIA's exposure to Thailand is less clear.
The US administration reportedly plans to introduce these new controls with the rollback of the more stringent “AI diffusion” export curb system introduced under the previous administration. If it goes through, this would mark the first major round of AI regulation under the new Trump administration since it scrapped the Biden-era rules. A similar proposal, the AI Diffusion Act, was introduced in January 2025 but later rescinded.
Without taking any single-name views, we make several observations for global investors with AI exposure:
New regulations, on the heels of the June rally, could trigger volatility. While the proposal appears relatively modest, we believe additional restrictions on AI chip exports could act still as a near-term overhang for cyclical tech stocks. Regulatory signals leading up to this have been relatively accommodative, with the White House reining in the Biden-era tiered AI diffusion rules. The recent rally in global tech was driven more by price-to-earnings multiple expansion than by positive earnings revisions, which suggests less room for any misses in second-quarter results. The upcoming earnings season also means the market is now in its peak blackout period for share buybacks, where companies are unable to step in and support share prices in the event of any declines.
US hyperscalers, data center operators could be shielded. The proposal reportedly includes a key concession: American data center operators will still be allowed to import US AI chips. This suggests major US hyperscalers, such as Google and Microsoft, can continue their regional data center expansions without significant disruption. We believe Malaysian companies primarily serving US-backed projects should remain relatively insulated from the new export restrictions. In contrast, data center operators building facilities speculatively or targeting Chinese demand may be more exposed to order cancelations or project delays.
We think AI capex demand, outlook remains solid. Our analysis shows that AI compute, including GPUs and custom AI chips, has attracted the strongest capital investment so far in 2025. This is reflected in robust spending commitments from the Big 4, sovereign buyers (including those in the Middle East), neoclouds, but also likely some China-linked demand. That said, the early April ban on H20 AI chip exports to China has already done much of the heavy lifting. We expect AI compute’s share of total AI capex to rise from approximately 53% in 2024 to about 59% in 2025. As investment broadens into areas like high-bandwidth memory, networking, and industrial AI infrastructure, we anticipate AI compute’s share will later normalize toward 50-55% from 2026 onward.
Against a backdrop of new potential chip restrictions and a broadening out of AI trends, we recommend investors adopt a balanced portfolio stance within tech and the AI value chain. Recently, we shifted our preferences away from some of the cyclical AI-linked stocks that rallied in June, tilting instead toward select AI laggards as well as internet and software stocks. Amid heightened volatility, we think investors may also want to consider structured strategies to help navigate market swings. Beyond near-term risks, we believe the robust secular trends in AI will continue to drive long-term gains in the tech sector. Data center stocks in Southeast Asia may be more sensitive in the near term to headline risk and potential policy changes. Within Chinese equities, we believe the impact on large-cap tech shares should be manageable.