UBS and Citi answer AI’s trillion-dollar question: can the tech reality match the current hype?

Will soaring demand for AI justify the massive infrastructure spending spree? UBS and Citi tackle the trillion-dollar question being asked by investors.

Nvidia chief executive Jensen Huang. Picture: AP

As artificial intelligence transforms from buzzword to business reality, two major investment brokers have tackled the trillion-dollar question: will demand for AI justify the massive spending spree that’s preceded it?

The answer, according to UBS and Citi, is a cautious yes – but with important caveats that investors need to understand.

UBS analyst Karl Keirstead and his team have been examining what they call the “three pillars” of AI demand and trying to work out whether the boom in AI infrastructure spending will actually pay off.

Their conclusion is, overall, reassuring for investors betting on the AI boom. However, they’ve identified a significant weak link in the chain.

Think of AI demand like a three-legged stool. The first leg is the tech giants building and training large language models; companies like OpenAI (which runs ChatGPT), Google, Meta and others racing to create smarter AI systems.

A Devodog robot; spread bets between ‘adopters’ and ‘enablers’. Photo: Thomas SAMSON / AFP

The second leg is consumer demand from people using ChatGPT, asking Google AI-powered questions, or interacting with AI chatbots. Both of these legs appear solid and growing stronger.

It’s the third leg – the enterprise demand – whereby Keirstead sees potential wobbles.

“Organisations are moving slowly, the return on investment is less clear, and AI technology needs to be architected to automate specific enterprise workflows and tasks,” the UBS report says.

Although consumers are embracing AI tools enthusiastically, businesses are being more cautious about spending big money on AI systems.

The risk is that enthusiasm from tech companies and consumers cools down before businesses fully embrace AI spending, creating a temporary “digestion phase” in the AI investment boom.

But UBS attaches “a low probability” to this scenario, saying it considers this as manageable rather than likely. The bank says it’s “constructive on AI demand trends,” arguing that training new AI models and growing consumer use will “sustain GPU demand for years to come”.

For investors wondering which stocks to back, Keirstead’s team has some clear favourites.

At the top of the list sits Nvidia, the chipmaker and stock market darling that’s become synonymous with AI processing power.

UBS also likes Broadcom as a key beneficiary of computing and networking demand, and Micron Technology for increased memory requirements.

Globally, it favours Taiwan Semiconductor.

UBS’s key AI stock picks:

• Hardware: Nvidia, Broadcom, Micron, TSMC, Arista, Ciena

• Software: Oracle, Snowflake, ServiceNow

• Tech giants: Meta

• Asian manufacturers: Quanta, Wistron

Interestingly, among software companies UBS likes the “infrastructure and data-exposed” names like Oracle and Snowflake, rather than the traditional software-as-a-service companies.

However, it thinks ServiceNow is better positioned among large SaaS players for AI monetisation.

Meanwhile, Citi US equity strategist Drew Pettit offers a complementary perspective, focusing on valuation and geographic opportunities. His key insight is that while American AI stocks generally trade at higher prices, they’ve “earned those multiples with strong growth plus higher and improving returns on equity and margins”. In investment language, the US offers “quality AI”.

But Pettit says the valuations of non-US AI stocks haven’t risen nearly as much, even as growth rates are expected to narrow. He therefore foresees “value AI” opportunities outside the US.

Citi’s analysis suggests that the growth expectations built into current AI stock prices are actually “attainable given consensus outlooks”.

Working backwards from current stock prices to see what growth rates are implied, the analysis finds that most AI stocks aren’t priced for impossibly high growth rates.

This suggests the AI rally isn’t built on completely unrealistic expectations.

Both banks are essentially saying that although AI stocks have risen substantially, the underlying business fundamentals can support these valuations if demand continues growing as expected.

Their analysis calls for a diversified approach to investing in the AI boom.

Rather than betting everything on the most obvious AI plays like Nvidia, they recommend spreading investments across what Citi calls “enablers” – companies that make AI possible like chip manufacturers; also “adopters” – companies which use AI to improve their businesses.

Overall, while the easy money to be made from AI investing may be over, both UBS and Citi expect sustainable demand drivers which can support continued investment returns.

The key is choosing companies with realistic valuations and genuine AI business models, rather than just riding the hype wave. Enterprise adoption continues to be the wildcard. If businesses embrace AI more slowly than expected, investors may face some bumpy periods ahead.

As always in investing, the devil is in the detail, and both reports say careful stock selection will matter more than simply buying any stock with “AI” in its business description.

Originally published as UBS and Citi answer AI’s trillion-dollar question: can the tech reality match the current hype?