The risks reshaping global AI investment

Video:The AI sell-off and US midterms: Live at the UBS Disruptive Tech Summit (6 mins)
CIO Live Replay: CIO Asia equity strategists on the AI sell-off and opportunities (30 mins)

NVIDIA’s shares fell more than 8% in after-hours trading on Wednesday, helping to pull Nasdaq futures around 1% lower and putting pressure on technology-heavy markets in Taiwan and South Korea. The moves followed reports that NVIDIA’s board had met on Wednesday night to discuss whether CEO Jensen Huang should remain in his role, following news that Huang had been the target of a foreign intelligence effort and subsequent developments.

These headlines come as investors are becoming more selective on the AI trade. Recent technology volatility reflects a shift from rewarding spending commitments toward assessing monetization, balance sheet strength, capacity utilization, and prospective returns. At the same time, underlying demand remains robust: Recent earnings have shown accelerating cloud revenue and expanding orders for computing capacity, and we forecast aggregate AI capital spending of USD 1.0tr in 2026 and USD 1.4tr in 2027.

Without taking any views on individual companies, we see several lessons that investors across the technology sector should consider:

Key-person risk warrants greater attention. The episode illustrates how leadership and governance can become material investment considerations when a company’s strategy, culture, and market standing are closely associated with an individual. This matters beyond any single company, particularly after several years in which AI-related equity gains have been concentrated in a relatively narrow group of businesses. As investors become more discerning about valuations and returns, governance and clarity around succession planning may play a larger role alongside earnings growth in determining performance.

Chip restrictions look unlikely to ease. AI competition between the US and China already operates within tight restrictions on exports of advanced semiconductors. We believe recent developments reduce the likelihood of an easing in current restrictions on advanced AI-chip shipments to China, while scrutiny of where processors are ultimately deployed could increase. Greater restrictions could weigh on sales in certain overseas markets, though they would not necessarily alter underlying AI demand, with compute capacity remaining in short supply.

AI is increasingly a geopolitical issue. As advanced computing becomes more strategically important, governments are paying closer attention to where chips are produced, sold, and deployed. This could result in greater tracking requirements, regulatory scrutiny, and friction around global AI supply chains, potentially raising costs and restricting access to some overseas markets. Greater geopolitical competition may also spur investment in alternative supply chains and domestic AI capacity, reinforcing the importance of distinguishing between strong underlying demand and the companies best positioned to capture it.

So, while we retain our conviction in the structural AI opportunity, we think these events reinforce the case for greater selectivity and diversification as the tech investment cycle evolves. We continue to favor financially strong hyperscalers; tighter future compute capacity could increase the value of existing infrastructure, while lower future capital spending could support free cash flow. Within AI “picks and shovels” (semiconductors and hardware), we would stay selective and favor quality companies with less margin expansion to unwind, alongside defensive technology exposures such as payment networks and potential beneficiaries of lower memory costs within consumer electronics. For investors whose technology exposure has become overly concentrated, we would consider broadening exposure within technology and diversifying across sectors, asset classes, and geographies.