- PwC expects global data center capex will hit $31.6 trillion by 2050 in a base case scenario, with spending accelerating over time
- Its forecast – and the lineup of winners and losers – hinges on power and chip availability
- Geopolitics and sovereignty regulations could also redraw the map for global data center investments
A new PwC forecast says global data center capex could hit anywhere from $22 trillion to $50 trillion by 2050. But the final number – and regional winners and losers – will be determined by a combination of power procurement and politics.
PwC’s forecast was based on modeling done by Oxford Economics and assumes a base case of $31.6 trillion in cumulative data center spending through 2050. The U.S. alone is expected to account for $15.1 trillion of that figure.
If the base case seems high, that’s because rather than hitting a peak and slowing down, PwC noted spending is actually expected to accelerate through the forecast period.
As the firm explained, bringing new data centers online marks the beginning – not the end – of an ongoing AI capex cycle. That’s because most of the spending is not going toward the facility itself but the equipment housed within it, which will need to be replaced every four to six years.
“What makes this capex cycle different is that annual spending accelerates over time, rising from roughly $800 billion in 2026 to $1.1 trillion in 2030 to $1.8 trillion in 2050,” PwC wrote. “Every prior infrastructure wave—railways, electrification, the internet—front-loaded construction capex and tailed off as the network matured. This wave inverts the pattern. The data center is essentially a chip-replacement subscription with a building wrapped around it.”
But there are a few key factors that will determine exactly when and where those subscriptions kick in.
What could impact the data center capex trajectory?
PwC highlighted two political maneuvers that could impact the overall capex trajectory as well as regional winners and losers. These include policy-related chip constraints and emphasis on sovereign solutions. And, thanks in large part to the U.S.' massive project pipeline and preponderance of AI technology leaders, the Americas have the most to gain or lose, depending on how the tide turns.
“The Americas have the largest absolute uplift if AI accelerates, with cumulative capex through 2050 rising to $27.1 trillion, and the largest absolute shortfall if it doesn’t, reflecting the GPU-intensive composition,” PwC wrote.
The rise of sovereignty-first regulations is one of the primary forces that could shape its fortunes – and those of developing countries.
While increased emphasis on sovereignty would only shave about $2 trillion off the base case capex scenario, it would significantly alter the regional breakdown of winners and losers.
“India, Vietnam, Indonesia, the Philippines, and Thailand all record material uplifts, reflecting large domestic demand bases that have so far been serviced disproportionately from regional hubs,” PwC wrote. The U.S., meanwhile, stands to lose around $2.9 trillion in capex in this scenario as projects are repatriated.
Silicon supply chain challenges
On the chip front, Circular Technology’s Brad Gastwirth recently noted that while global demand remains strong for technology like Nvidia’s GPUs, questions are already starting to arise around supply.
“We continue to believe the important question is shifting away from whether AI infrastructure demand remains strong toward whether the supply chain can expand quickly enough to support it,” he wrote follow Nvidia’s recent earnings call.
The role of geopolitics
As far as politics goes, a tit-for-tat battle between the U.S. and China has already limited proliferation of the former’s technology in the latter country. If the trade war between the two countries intensifies such that advanced GPUs are harder to get in more markets and raw materials are harder to source along the semiconductor supply chain, the consequences could be dire.
“The headline impact is severe in the near term and partially recovers over time. Annual capex falls to roughly half the central scenario by 2030, before recovering as supply chains adapt,” PwC wrote. It added the Middle East and China would be among the most heavily impacted, with the Americas taking the biggest hit in dollars thanks in large part to the U.S.’ GPU-dependent project pipeline.
The role of the power supply
But more than any politicking, power supply will determine where data center capex is spent, PwC said.
“Power sits at the top of the list because affordable, reliable, and increasingly low-carbon electricity at scale is the hardest requirement for many markets to meet—and delivering it quickly is harder still,” PwC wrote.
In April, the International Energy Agency updated its data center forecast, predicting electricity consumption in the sector will roughly doubly from 485 TWh in 2025 to 950TWh in 2030. AI data center consumption specifically is expected to grow faster than the sector overall, with electricity usage tripling.
Goldman Sachs tipped power demand to grow even faster – jumping 2.7x from 2025 to 2030. Put another way, the data center sector is expected to add “the power equivalent of the whole country of Japan, which is the number five power consuming country” in the seven-year period from 2024 to 2030, Goldman Sachs Research’s Brian Singer said.
This skyrocketing power demand is running into a range of deployment constraints, everything from permitting and parts (like transformers and turbines) to people (think skilled workers like electricians) and pricing, Goldman Sachs’ team noted. And in certain markets, the supply-demand equation is already “critically tight,” Goldman’s Carly Davenport said.
Put it all together and PwC noted that “The operators that can secure megawatts faster than competitors will capture a disproportionate share of capacity regardless of which path the market takes.”
Read more about the data center buildout:
Data centers become ‘sleeper issue’ that could decide elections
Ban on data center transceiver imports would be ‘terrible,’ analysts say
West Virginia joins cadre of states enacting data center development plans
About 700 data centers are being built in U.S. as public outcry increases
U.S. Anthropic order throws a megaton of fuel on sovereign AI fire
Facts Only
* PwC forecasts global data center capex will reach $31.6 trillion by 2050 in a base case scenario.
* The forecast range for global data center capex by 2050 is between $22 trillion and $50 trillion.
* The U.S. alone is expected to account for $15.1 trillion of the base case figure.
* Spending is expected to accelerate over the forecast period, not peak and slow down.
* Annual spending is projected to rise from roughly $800 billion in 2026 to $1.1 trillion in 2030 to $1.8 trillion in 2050.
* Data center capex is viewed as an ongoing AI cycle where spending focuses on equipment replacement every four to six years.
* The Americas have the largest absolute uplift if AI accelerates, with cumulative capex projected at $27.1 trillion through 2050.
* Sovereignty-first regulations could reduce the base case capex by approximately $2 trillion.
* India, Vietnam, Indonesia, the Philippines, and Thailand record material uplifts due to domestic demand bases.
* The U.S. stands to lose around $2.9 trillion in capex if projects are repatriated due to sovereignty concerns.
* Silicon supply chain challenges raise questions about the ability of supply chains to expand quickly enough for AI infrastructure demand.
* The Middle East and China are expected to be among the most heavily impacted by geopolitical semiconductor constraints.
* Power availability is determined to be a key factor in data center capex spending.
* Data center electricity consumption is expected to roughly double from 485 TWh in 2025 to 950 TWh in 2030.
Executive Summary
Global data center capital expenditure is forecasted to range between $22 trillion and $50 trillion by 2050, with a base case scenario from PwC modeling $31.6 trillion in cumulative spending through that year. This forecast is contingent upon the availability of power and chips, as well as geopolitical factors. The acceleration in spending is driven by the fact that building data centers represents the beginning of an ongoing AI capex cycle, where most expenditure is on equipment replacement rather than facility construction. Annual spending is projected to accelerate from approximately $800 billion in 2026 to $1.8 trillion in 2050.
The trajectory of investment across regions will be heavily influenced by political maneuvers and supply chain constraints. The Americas hold the largest potential gain or loss based on AI acceleration, reflecting their concentration of GPU technology leaders. Furthermore, the rise of sovereignty-first regulations is expected to redistribute investment flows, potentially shifting capital away from areas like the U.S. towards regions such as India, Vietnam, Indonesia, the Philippines, and Thailand, which show material uplifts due to domestic demand.
Power supply remains a critical constraint, as affordable, reliable, and low-carbon electricity at scale is a requirement for deployment across various markets. The demand for power in the data center sector is expected to grow significantly faster than general energy consumption, with some projections indicating that this sector could add the power equivalent of a major consuming nation over the next seven years. Ultimately, the capacity to secure power supply rapidly will determine which operators capture market share, irrespective of the overall investment path.
Full Take
The narrative surrounding data center investment shifts from a linear construction phase to an accelerating subscription model, where the value resides in continuous equipment refreshment rather than initial build-out. This inversion of historical infrastructure spending patterns suggests that future investment will be dictated more by technological iteration—specifically AI chip requirements—than by traditional development cycles. The core tension is between physical reality (power and material supply) and geopolitical friction.
The implication of sovereignty-first regulation acts as a powerful redistributive force. While the overall capital pool might be constrained, regional divergence is inevitable, rewarding nations with large domestic demand bases. This points toward a future where investment geography is determined less by purely economic incentives and more by political alignment and control over critical inputs. The competition over power—which is framed as the hardest requirement for deployment—introduces an entirely new bottleneck that supersedes raw capital flow concerns.
The dynamic between US and China semiconductor rivalry acts as a multiplier on geopolitical risk, suggesting that supply chain fragility will translate directly into capex volatility in specific regions. If resource control tightens, the acceleration seen in some areas may be offset by near-term slowdowns, especially where critical inputs are contested. The ultimate locus of future investment capacity depends not just on market demand or technological appetite, but on the ability of regional actors to establish autonomous supply chains and secure energy infrastructure outside dominant geopolitical spheres.
Bridge Questions: If sovereignty mandates lead to a significant shift in capital allocation, what specific governance frameworks would be most effective at mitigating short-term economic contraction while promoting long-term regional resilience? How will the accelerating demand for power fundamentally alter international energy diplomacy over the next decade? What mechanisms could reconcile the need for rapid deployment with the necessity of securing geographically distributed supply chains?
Sentinel — Human
The text functions as a high-level aggregation of expert forecasts regarding data center expansion, effectively synthesizing complex quantitative and geopolitical factors into an analytical narrative.