New phase of AI boom

THE artificial intelligence (AI) investment cycle looks set to run further, with global spending potentially reaching levels comparable with previous major technology buildouts.

Investors are likely to become more selective, however, as attention shifts from simply funding AI infrastructure to identifying companies that can turn the spending boom into sustainable profits.

Goldman Sachs Research estimates that AI-related investment globally will reach about US$1 trillion in 2026, including US$581bil in the United States, as spending extends well beyond the large cloud companies traditionally used to gauge the scale of the boom.

The estimate suggests the widely cited forecast of about US$794bil in capital expenditure (capex) by US hyperscalers understates global AI investment by roughly US$200bil, while overstating the amount of US investment by a similar amount.

The broader measure includes spending by other public companies, private companies and AI-exposed businesses outside the United States, including in Asia.

That matters for investors because the scale of the AI buildout is larger and more geographically diverse than the headline hyperscaler numbers suggest.

The research also points to further expansion ahead. AI capex is estimated to rise from 1.8% of US gross domestic product (GDP), with global AI investment at 0.9% of global GDP, in 2026 to 2.5% of US GDP and 1.3% globally in 2027.

By 2028, those figures are likely to rise further to 2.8% and 1.4%, respectively.

“The AI capex growth outlook, including how high AI investment ultimately rises as a share of GDP and when capex growth slows, is a key source of uncertainty for macro markets right now,” Joseph Briggs, co-leader of the Global Economics team, writes.

Robust near-term outlook

For investors, the key issue is therefore not simply whether AI spending continues, but how long the investment cycle can maintain its current pace and which parts of the technology ecosystem ultimately capture the returns.

Goldman Sachs Research says its estimates point to a cumulative US$1.8 trillion of AI investment by end-2026.

It also finds that two separate methods used to cross-check its estimates produce broadly similar results, with both indicating global AI investment of around US$1 trillion this year and just under US$600bil in the United States.

The research uses a range of indicators to assess whether a slowdown in AI capital expenditure is approaching, including semiconductor manufacturing equipment imports in Taiwan and South Korea, relevant Purchasing Managers’ Index indicators, import prices, memory purchase prices and graphics processing unit rental prices.

“The good news for the capex outlook is that all leading indicators rank near the top end of their range since 2022,” Briggs writes. “This pattern suggests a robust near-term growth outlook.”

That provides some support for the continuation of the investment cycle in the near term, although the size of the spending boom also raises the stakes for markets.

“These levels are consistent with the 2% to 5% of GDP peak investment impulses observed in prior general-purpose technology buildouts,” Briggs writes.

The investment opportunity is, meanwhile, becoming more differentiated across the technology sector.

Future profits the focus

Franklin Templeton Research says investors are not abandoning AI following the recent weakness in semiconductor shares.

Instead, capital is rotating within the AI ecosystem as investors assess which companies are best positioned to generate future profits.

It divides the opportunity into three broad areas: AI infrastructure, AI platforms and AI applications.

Infrastructure includes chips, networking equipment, power systems and data centres, while platforms include large cloud companies such as Microsoft Corp, Amazon.com Inc, Alphabet Inc and Meta Platforms Inc.

Applications cover software companies developing AI-powered products and services.

Semiconductors remain central to the investment cycle because every AI model requires advanced chips, memory, networking equipment and servers.

Large cloud providers continue to invest heavily in infrastructure, supporting demand across the semiconductor supply chain.

But the sharp gains in semiconductor shares have also made investors more sensitive to valuations and the possibility that spending growth could eventually moderate.

Franklin Templeton Research says the recent pullback appears to reflect concerns over expectations rather than a loss of confidence in AI demand.

The focus is increasingly moving towards the large US hyperscalers, which occupy a different position because they both fund AI infrastructure and own the platforms through which AI services are delivered.

Potential sources of revenue include cloud services, enterprise software, digital advertising, consumer platforms, AI assistants and productivity tools.

This creates a distinction between companies supplying the tools for AI and those positioned to monetise its wider adoption.

Some investors are consequently shifting exposure from semiconductor companies towards large platform businesses, according to Franklin Templeton Research.

The software segment is also attracting a more selective approach.

Many software shares have fallen over the past year as investors worried that AI could make existing products less valuable or increase competition.

However, Franklin Templeton Research points out that many enterprise software platforms handle critical functions such as financial records, supply chains, customer data, human resources and security.

These systems are often deeply embedded within businesses, making them difficult and costly to replace.

In those cases, AI could potentially enhance rather than undermine their value by making existing platforms more useful and productive.

“The recent market rotation should not be viewed as a move away from AI. Rather, we believe it reflects a shift in where investors believe the biggest future opportunities may lie.”

That shift also highlights a growing portfolio consideration: concentration risk.

Addressing concentration risk

Charles Schwab Corp notes that the technology sector now represents an unusually large share of global equity indices and is expected to account for more than half of global equity market earnings growth in 2026, based on consensus estimates and its calculations.

While a rising index weight can benefit investors as capital flows towards companies with strong fundamental growth, it can also increase exposure to a single source of returns.

“From our perspective, the combination of market cap and earnings concentration – and the circularity of these conditions (including the large reliance on the AI theme) – is where the risks add up,” the financial group points out.

Charles Schwab does not see this as an indication that the AI investment cycle is about to end.

Instead, it highlights the range of possible outcomes at a time when technology valuations and earnings expectations already reflect a fairly optimistic scenario.

“These considerations do not mean we foresee an imminent end to the AI investment cycle or the powerful earnings growth it’s generating,” it says.

The implication for investment strategy is therefore more about positioning than retreating.

AI infrastructure remains central to the spending cycle, while hyperscalers and selected software companies may offer greater exposure to the revenue and profit opportunities that emerge as adoption broadens.

At the same time, the concentration of market capitalisation, earnings and expectations around technology means investors may need to look beyond the most obvious beneficiaries.

“While we can’t predict how the AI innovation cycle will play out, we believe more active diversification can potentially improve risk-adjusted returns across a range of potential scenarios,” Charles Schwab Corp notes.

“This isn’t a call to abandon the tech sector or those companies most benefitting from the related investment cycle, but we do believe more actively diversifying portfolios is warranted,” it adds.