CLSA says AI infrastructure spending will stay high, chip shortage to remain key constraint

Resumen detallado
Bhavtosh Vajpayee, International Head of Technology Research at CLSA, expects AI infrastructure spending to continue through the rest of the decade and argues that chip supply, rather than data centre availability, is the real constraint. He also named Taiwan Semiconductor (TSMC) as his top AI investment pick.
The recent correction in global AI-related stocks could present a buying opportunity as the artificial intelligence infrastructure cycle is far from over, according to Bhavtosh Vajpayee, International Head of Technology Research at CLSA.
Speaking at the CITIC CLSA India Forum, Vajpayee said the AI build-out is expected to continue through the rest of the decade despite investor concerns over funding, returns and policy changes.
CLSA expects AI infrastructure spending to remain elevated for several years as hyperscalers continue investing to protect their market position. Vajpayee also said Taiwan Semiconductor Manufacturing Co. (TSMC) remains the firm's preferred way to play the AI theme, with its 2027 capital expenditure guidance likely to shape expectations for the broader AI ecosystem.
"The build is robust and will continue through the rest of the decade," Vajpayee said, adding that recent weakness reflects "a degree of tiredness, exhaustion among investors" rather than a structural change in the AI story.
He said concerns surrounding Nvidia, delays in the Anthropic IPO and questions around AI returns have weighed on sentiment, but such corrections are common during major technology investment cycles.
Vajpayee said investors expecting hyperscalers such as Microsoft, Google and Meta to quickly return to strong free cash flow generation may have to wait longer.
"It is a reality of the new AI build that more money has to be spent," he said, explaining that companies are prioritising AI infrastructure investments to maintain their competitive position.
According to him, capital expenditure intensity has structurally increased and is unlikely to moderate for several years because hyperscalers do not want to lose leadership in AI.
Over time, however, he expects AI monetisation to improve as enterprise adoption expands.
Inicio del artículo, extracto de CNBC-TV18. Para leer el resto, abra el artículo en el sitio del medio.
MarketNews muestra el titular y el resumen publicados por el medio. El artículo completo se lee en el sitio del editor. (www.cnbctv18.com)



