Data center investment could hit nearly $1.6 trillion by 2030 under forecasts by research firm Omdia, with demand for AI compute capacity continuing to grow, despite lagging AI adoption and the potential of an AI bubble.

Omdia’s December Cloud and Data Center Market Snapshot, published by The Register, shows that, in Omdia’s likeliest forecast scenario, data center capex would reach $1.6 trillion in 2030. This scenario is the firm’s “consensus” scenario.

Omdia said this scenario “considers actual order pipeline and demand which are both strong.

“This is balanced against numerous constraints (power availability, manufacturing capacity, supply chain hiccups),” the company added.

Omdia explained that the scenario was aligned to Nvidia’s 2025-2026 order backlog.

The research firm considered a number of different scenarios before coming to its consensus, including one that referenced fears of an AI bubble.

The “bubble scenario,” which Omdia only considers to be a five percent likelihood, inexplicably forecasts data center capex growing faster than in all other scenarios, at $1.4 trillion in 2027 compared to the $1.1 trillion in 2027 forecast in its likeliest scenario.

Omdia then forecasts a drop in data center capex after 2027 in the “bubble scenario,” representing the bubble bursting and bringing capex down to just more than $1 trillion in 2028. At its lowest point, post-burst, in 2028, Omdia still expects data center spend to eclipse 2025 spend, which currently sits at around $700 billion, according to the firm’s data. The forecast then projects data center capex regaining momentum through 2030.

The research firm said this scenario reflects “a failure to realize productivity gains through AI use quickly enough, and after five years of accelerated investment, investors get spooked.”

“We consider 2027 to be a key year because key AI developers have made revenue commitments for that year.”

In its report, Omdia said that demand for AI compute capacity is strong, but supply constraints are hampering adoption efforts. In coming to its consensus, the research firm said “more users and higher usage per user is coming,” though it provided no rationale for this claim.

In recent weeks, fears of overspending on AI compute capacity have grown, with significant doubts emerging over the profitability of such projects.

IBM’s CEO, Arvind Krishna, said last week that there was “no way” companies chasing AI would get returns on commitments to multi-gigawatt data center capacity.

Krishna explained: “If you are going to commit 20 to 30GW, that's one company, that's $1.5 trillion of capex. If I look at the total commits in the world in this space, in chasing AGI, it seems to be like 100GW with these announcements," which would cost around $8 trillion.

He added: "It's my view that there's no way you're going to get a return on that, because $8 trillion of capex means you need roughly $800 billion of profit just to pay for the interest."

There’s also plenty of discourse over the potential for overcapacity in the industry, as the gold rush of investment, combined with the uncertainty of demand could mean that current buildout will outpace the development of practical and revenue-generating AI applications.

Addressing some of these concerns, in its consensus scenario, Omdia suggested data center buildout is occurring more slowly than announcements, reducing the chance of “overbuild,” though the referenced report did not provide any data to support this.