Is the Data Center Market growing now?

In our data center market deck, you will find everything you need to understand the market
SUMMARY
Yes. The Data Center Market is growing very fast now, and the clearest proof is that new physical capacity is being delivered, occupied and contracted at the same time.
The strongest evidence is the match between supply and demand. In North America, roughly 2.5 GW of new capacity was added in 2025 and net absorption was also about 2.5 GW, while vacancy still fell to a record low.
AI is the biggest accelerator of new demand, but it is not the whole market. Hyperscalers still run cloud, storage, databases, enterprise software and consumer services, so AI can dominate incremental power requirements without accounting for most workloads already inside data centers.
The ownership story is more mixed than the usual “Big Tech builds everything itself” narrative. Hyperscalers are expanding private campuses and leasing huge blocks from commercial operators at the same time because their capacity needs are simply too large to satisfy through one model.
The construction pipeline looks unusually credible because so much of it is already spoken for. JLL tracks more than 66 GW under construction in North America, with around 95% precommitted, so the near-term boom is not just a stack of speculative announcements.
Power has become a bigger constraint than customer demand in many markets. The scarce product is increasingly not a building, but a usable block of high-density capacity with enough electricity, cooling, fiber and a realistic connection date.
Rents and vacancy show that supply growth has not yet broken pricing power in the tightest hubs. Even after large capacity additions, Northern Virginia, Atlanta and Singapore remain extremely constrained, while older leases are often repricing upward.
The financial risk sits less in whether money is available today and more in whether AI economics eventually justify the scale of infrastructure spending. Capital markets are still funding the buildout, but hyperscaler CapEx has become large enough to pressure free cash flow and raise the cost of being wrong.
Electricity consumption gives an independent check on the real-estate data. Data centers already account for around half of recent U.S. electricity-demand growth, and both the IEA and Lawrence Berkeley expect their share of power consumption to rise sharply through 2030.
The main weak spots are local rather than market-wide. Querétaro has already shown how vacancy can jump after a huge supply wave, some Chinese facilities have struggled with utilization, and power or permitting delays can stop otherwise viable projects.
Through 2030, the direction still looks much clearer than the exact size. Capacity, contracted demand, hyperscaler spending and electricity forecasts all point upward, but today’s most extreme growth rates will probably become more uneven as power, financing and local opposition decide which projects actually open.

This market map, featured in our data center market deck, highlights top companies and startups in the data center market
Is the Data Center Market Growing Now?
What do we actually mean by the Data Center Market?
The Data Center Market is growing if companies are adding and using more physical computing capacity, whether that capacity sits inside a hyperscaler-owned campus or a commercial colocation facility.
That definition keeps us away from a surprisingly common problem: market-size estimates for data centers often measure different things. Some include buildings and power infrastructure. Others add servers, networking equipment, managed services or pieces of the cloud market. Two research firms can therefore publish very different dollar figures without actually disagreeing about what is happening on the ground.
We get a cleaner answer by following physical capacity in megawatts and gigawatts, net absorption, vacancy, rents, signed leases, operator revenue, hyperscaler infrastructure spending and electricity use.
The ownership mix also matters. Synergy Research Group counted 1,360 large hyperscale data centers at the end of 2025 and estimated that hyperscalers controlled 48% of all worldwide data-center capacity. Nearly 60% of that hyperscale capacity was in facilities they owned themselves, while the rest was leased. Traditional colocation accounted for another 20% of global capacity.
So an Amazon-built campus and an Equinix facility both belong in this analysis. They simply represent different parts of the same physical infrastructure market.
Is the Data Center Market actually adding a lot of capacity now?
Yes, the Data Center Market is adding physical capacity extremely fast right now, with double-digit growth across every major region tracked by CBRE.
CBRE measured roughly 16 GW of inventory across 16 large global data-center markets in the first quarter of 2026, up 25% in one year. That represents roughly 3.2 GW of additional operating inventory across those markets alone.
North America grew fastest among the mature regions. Inventory across Northern Virginia, Atlanta, Dallas-Fort Worth and Chicago jumped 33%. Europe was more surprising: London, Frankfurt, Paris and Amsterdam grew 18.9%, compared with only 7.2% during the previous comparable period. Tokyo, Singapore, Sydney and Hong Kong collectively grew 13.4%.
Latin America came from a smaller base but expanded 41.3%. Querétaro was the extreme case, with inventory jumping more than fivefold after several large hyperscale deployments.
Northern Virginia alone added 1,135.9 MW in twelve months and reached 4,182 MW. Dallas-Fort Worth added almost 380 MW. Frankfurt added enough capacity to take its total above 1.2 GW. Tokyo passed 1 GW.
These are useful numbers because CBRE is measuring delivered inventory rather than every campus somebody has announced for 2029.
| Market group | YoY inventory growth | What changed |
|---|---|---|
| Major North American markets | 33% | More than 1 GW added in Northern Virginia alone |
| Major European markets | 18.9% | Growth accelerated sharply from 7.2% previously |
| Major Asia-Pacific markets | 13.4% | Tokyo passed 1 GW despite difficult power markets |
| Major Latin American markets | 41.3% | Querétaro was transformed by hyperscale deployments |

As this chart shows, and as featured in our data center market deck, search interest in data centers has increased significantly
Is demand for data centers keeping up with all that new supply?
Yes, data-center demand is currently swallowing new supply about as quickly as developers can deliver it.
The best comparison we found comes from CBRE's eight primary North American markets in 2025. Total supply increased 36% to 9,432 MW. That means roughly 2.5 GW of net capacity was added during the year.
Net absorption was 2,497.6 MW.
Those numbers are almost identical. In simple terms, the major North American markets added around 2.5 GW and users moved into around 2.5 GW during the same year. Vacancy still fell to a record 1.4%.
The pattern remained very strong in the first quarter of 2026. Across Northern Virginia, Atlanta, Dallas-Fort Worth and Chicago, CBRE recorded 2,236 MW of net absorption, 34% more than a year earlier.
Northern Virginia is particularly revealing. Inventory increased by 1,135.9 MW over twelve months while net absorption reached 1,148.3 MW. Even after adding more than a gigawatt of capacity, vacancy fell from 0.8% to 0.3%.
JLL's broader North American dataset captures a much larger hyperscale universe and points in the same direction. Its latest midyear report measured 25 GW of absorption during the first half of 2026, twice the level a year earlier and about five times what it recorded two years earlier.
Is Data Center Market growth mostly an American story?
No, Data Center Market growth is global today, although the United States has pulled unusually far ahead in the hyperscale part of the market.
Synergy Research Group estimates that 15 of the world's 20 largest hyperscale locations are now in the United States. The U.S. controlled around 55% of worldwide hyperscale operational capacity in late 2025, up from roughly 52% three years earlier.
That concentration is remarkable, but the rest of the world is still adding capacity quickly.
Europe's four largest markets grew inventory by 18.9% in CBRE's latest global report. Frankfurt expanded 23%, Paris almost 15%, and London added enough new capacity to grow around 21% despite severe power constraints around the city.
Tokyo now has more than 1 GW of inventory. Singapore remains so tight that vacancy sits around 2%, and Digital Realty has most recently been selected to develop another 50 MW of capacity there. Restrictions in Singapore have also helped nearby Johor emerge as a much larger data-center hub.
Latin America is scaling faster from a smaller base. São Paulo remains the region's main hub, while Querétaro reached almost 300 MW after its latest wave of hyperscale construction.
China is building heavily too. Synergy already ranks Greater Beijing among the world's two largest hyperscale markets. Chinese policy is increasingly steering new computing infrastructure toward large national hubs with better access to land and electricity after an earlier construction wave left some smaller facilities poorly utilized.

This chart, featured in our data center market deck, illustrates yearly venture capital funding for data center startups
Is AI really driving the Data Center Market, or is that exaggerated?
AI is clearly the biggest new accelerator of the Data Center Market, but cloud computing, enterprise workloads and ordinary digital services are still creating substantial demand.
JLL's latest North American demand breakdown is useful here. Hyperscalers represent 59% of current demand, neocloud providers 11% and pure-play AI companies another 7%. Enterprise customers are still competing for capacity too, often seeking blocks between 500 kW and 3 MW.
The hyperscaler category itself cannot be treated as synonymous with AI. AWS runs databases, storage, enterprise applications, streaming infrastructure, ecommerce systems and thousands of other workloads alongside AI.
Amazon's latest results make the distinction obvious. AWS revenue grew 37% year over year to $42.2 billion for the quarter, its fastest growth in 18 quarters and a $169 billion annualized revenue run rate. AI contributed heavily to that acceleration, but AWS was already one of the world's largest infrastructure businesses before the generative-AI boom.
Google Cloud grew even faster, with second-quarter revenue jumping 82% to $24.8 billion. Alphabet said enterprise AI infrastructure and AI solutions were major drivers, alongside its broader Google Cloud Platform business.
Equinix gives us another view. The company added a record 9,700 net interconnections in its latest quarter. Those connections cover cloud platforms, financial firms, telecom networks, enterprises and AI systems exchanging data with one another.
AI has changed the size of individual requirements most dramatically. A conventional enterprise customer may ask for a few megawatts. AI training projects can demand hundreds of megawatts, which means AI can dominate incremental power demand without accounting for every workload already running inside data centers.
If you want more recent data on this point, please see our latest data center market report.
Are Amazon, Google and Microsoft just building their own data centers instead of using operators like Equinix?
No, hyperscalers are building huge private campuses and leasing huge amounts of commercial data-center capacity at the same time.
Synergy estimates that hyperscalers now control 48% of worldwide data-center capacity. Nearly 60% of their capacity sits in facilities they own, leaving more than 40% in leased facilities.
Using those shares, hyperscaler-owned sites represent roughly 29% of worldwide data-center capacity, while hyperscalers occupying leased facilities account for roughly another 19%.
That balance is still shifting toward self-build. Synergy expects hyperscalers to control 67% of total capacity by 2031. Yet the same forecast has non-hyperscale colocation capacity continuing to grow at close to double-digit annual rates.
The reason is fairly straightforward. Amazon, Microsoft, Google and Meta need so much capacity that they cannot satisfy every requirement with facilities they develop themselves. They can own strategic campuses while leasing elsewhere for faster delivery, geographic coverage or additional capacity.
JLL currently attributes 59% of North American tenant demand to hyperscalers while simultaneously tracking tens of gigawatts of owner-occupied hyperscale construction.

This chart, featured in our data center market deck, shows how Equinix is capturing share in data centers
Are Big Tech companies still spending more on data centers now?
Yes, Big Tech infrastructure spending is still going up aggressively, and several hyperscalers have recently raised their plans again.
Amazon now expects roughly $220 billion of capital expenditure in 2026 after increasing its plan by about 10%. CEO Andy Jassy said the company still lacked enough compute capacity to satisfy demand despite spending at that level.
Microsoft has said it expects approximately $190 billion of calendar-year capital expenditure. Management also expects the company to remain capacity constrained through at least 2026 even as it brings GPUs, CPUs, storage and additional data-center capacity online.
Alphabet recently raised its 2026 guidance from $180 billion-$190 billion to $195 billion-$205 billion. The company says roughly 60% of current capital spending is going toward servers, with much of the remaining 40% going into data centers and networking.
Meta has increased the bottom end of its annual range and now expects $130 billion-$145 billion.
At the midpoint of Alphabet and Meta's ranges, those four companies alone are planning about $748 billion of capital expenditure this year. That works out to a little over $2 billion per day.
We should not call all $748 billion "data-center spending." Chips, servers, networking equipment and other infrastructure sit inside those budgets. Still, the companies themselves repeatedly identify AI compute and infrastructure capacity as the reason spending has moved so much higher.
| Company | Current 2026 CapEx plan | Recent business evidence |
|---|---|---|
| Amazon | About $220B | AWS revenue +37%; Amazon says capacity remains insufficient |
| Microsoft | About $190B | Management expects compute constraints to continue through 2026 |
| Alphabet | $195B-$205B | Google Cloud revenue +82%; CapEx guidance raised again |
| Meta | $130B-$145B | Lower end of annual spending guidance raised |
If you want more recent data on this point, please see our latest data center market report.
Are actual data-center companies growing too?
Yes, commercial data-center operators are growing today, and their bookings suggest more revenue is already waiting to come online.
Digital Realty reported $1.9 billion of second-quarter revenue, 29% more than a year earlier. Adjusted EBITDA increased 19%.
The leasing numbers are more useful than the headline revenue figure because unusual items affected the quarter. Digital Realty signed $307 million of annualized GAAP base-rent bookings at 100% ownership share during the quarter. Shortly afterward, it signed two additional hyperscale leases representing another $410 million of annualized base rent at 100% share.
Its signed-but-not-started lease backlog reached a record $1.9 billion of annualized base rent at 100% share. That backlog represents customers who have already signed but are still waiting for their contracted capacity to begin.
Equinix is seeing growth in a different part of the market. Monthly recurring revenue increased 11% year over year in its latest quarter. Annualized gross bookings reached $424 million, up 23%, while the company added a record 9,700 net interconnections.
Equinix also had 52 expansion projects underway across 33 markets after adding nine projects since its previous update.
| Operator | Latest result | What we learn |
|---|---|---|
| Digital Realty | Revenue +29% YoY | More demand is reaching commercial landlords |
| Digital Realty | $1.9B signed lease backlog | A large amount of future rent is already contracted |
| Equinix | Monthly recurring revenue +11% | Existing platform revenue is still expanding |
| Equinix | Gross bookings +23% | New customer commitments remain strong |

This chart, featured in our data center market deck, illustrates yearly funding for data center startups
Are data-center rents still going up?
Yes, data-center rents are still rising in many major markets even after a huge wave of new construction.
CBRE says the average asking rate for a 250-to-500-kW requirement across its primary North American markets rose 6.5% in 2025 to about $195 per kW per month. That was the fourth consecutive annual increase.
The latest global numbers show substantial differences by location. Chicago was around $200-$230 per kW per month for that size of requirement. Northern Virginia ranged from roughly $190 to $235. Frankfurt was around $235-$265.
Singapore sits in another league. CBRE currently estimates average asking rents around $403 per kW per month because the city tightly controls new capacity and has very little available space.
Existing leases are repricing upward too. Digital Realty reported a 25.4% cash increase on renewal leases signed during its latest quarter. That figure will move around considerably from quarter to quarter, so we would not use it as a general market-rent index. It does show the pricing power available when older contracts reset in an extremely tight market.
Is the giant data-center construction pipeline actually real?
A large part of the near-term data-center pipeline is already contracted, which makes today's construction boom much more credible than the raw announcement numbers suggest.
JLL currently tracks more than 66 GW of capacity under construction in North America. Around 95% is precommitted.
That leaves only about 3.3 GW uncommitted across a 66 GW pipeline.
The ratio matters more than the headline number. Developers are certainly planning projects beyond the construction stage that may never happen, especially when power connections stretch years into the future. Capacity already under construction looks much less speculative when customers have reserved almost all of it.
JLL says many companies signing contracts today are already securing capacity for 2028 delivery.
CBRE's narrower commercial-market data shows something similar. In Dallas-Fort Worth, for example, 88% of the 716.7 MW under construction in its latest global report was already preleased.
Contract quality still matters. A twenty-year commitment from a huge investment-grade cloud company is different from a lease signed by a young AI infrastructure company burning cash. Some projects will also be delayed by substations, transmission lines, generators or permitting.
| Stage | How seriously we should take it | Why |
|---|---|---|
| Operating capacity | Very seriously | Customers can already use it |
| Under construction and precommitted | Seriously | Capital is being spent and customers have signed |
| Under construction but unleased | More cautiously | Demand still needs to materialize |
| Announced or planned campus | Very cautiously | Power, permits, financing or demand may still kill it |

This chart, featured in our data center market deck, compares the main business model options for hyperscale data center operators
How can data centers be in shortage when some markets already have empty space?
Data-center shortages are real because the capacity customers need today is highly specific, while empty capacity can sit in the wrong building, city or power configuration.
Northern Virginia had only 0.3% vacancy in CBRE's latest global comparison. Atlanta was at 1%. Singapore was around 2%.
Querétaro tells a different story. Its inventory jumped 450% in a year, and vacancy rose from 0.9% to 10.6% as an extraordinary amount of new capacity entered the market. London also saw vacancy move from 8% to 8.6% after large additions.
Technical differences add another layer. A legacy facility designed for 5 or 10 kW racks cannot automatically host a modern GPU cluster demanding 40 kW, 80 kW or considerably more per rack. Cooling, electrical distribution and floor design can all become limiting factors.
China has already shown what badly matched supply can look like. After an earlier subsidy-driven expansion, authorities have been pushing for better coordination of computing resources and discouraging some smaller developments. Facilities can exist physically while still being unattractive to the customers creating the strongest demand.
The scarce unit in the current Data Center Market is usually a usable block of high-quality powered capacity: enough electricity, appropriate cooling, strong fiber, the required rack density and a location that works for the workload.
If you want more recent data on this point, please see our latest data center market report.
Is electricity becoming the biggest problem for Data Center Market growth?
Yes, electricity has become the main bottleneck for Data Center Market growth in many locations, and the latest developments show how quickly power limits can override customer demand.
CBRE says access to 300 MW or more within a usable timeframe has overtaken pure fiber connectivity as the top site-selection issue for many large projects. Traditional 12-to-18-month development cycles make little sense for 500-MW AI campuses that require new substations, transmission upgrades or power generation.
Chicago gives us one example. Some ComEd delivery timelines now extend to 2032 or later. West London's key substation upgrades are also expected only in the early 2030s. Frankfurt faces similar transmission constraints.
Those delays are reshaping the map. JLL says 77% of North American capacity currently under construction sits in what it calls frontier markets. West Texas, Ohio, Louisiana, Indiana and the Carolinas have gained projects that might previously have gone to Northern Virginia, Silicon Valley or another established hub.
Texas shows how abruptly the constraint can become political. After the state's governor paused some new data-center development to examine projects under review, the U.S. Energy Information Administration cut its forecast for Texas electricity-load growth in 2027 from 14% to 6%.
The customers did not disappear between those forecasts. The amount of demand that the grid might actually connect changed.
Community opposition is becoming part of the same bottleneck. JLL's latest U.S. survey found that 79% of respondents support American leadership in AI, while only 14% support data-center development in their own community. Electricity bills, water, land use and noise can turn an economically attractive project into a politically difficult one.

This chart, featured in our data center market deck, shows the revenue mix across customer segments in the data center market
Is there enough money to finance the data-center boom, and could weak AI economics derail it?
Yes, capital is still readily available for data centers today, but the amount being spent is now large enough that weak AI returns could eventually slow the market.
JLL estimates that projects currently being built could create more than $700 billion of permanent debt originations over the next 30 months.
AI-related bond issuance reached around $250 billion in the first half of 2026 across investment-grade and high-yield markets. Financing markets are therefore supporting the buildout rather than shutting it down.
The cost of each project explains why capital has become such an important part of the story. JLL estimates average shell-and-core construction costs rose from around $7.7 million per MW in 2020 to $10.7 million in 2025 and could reach roughly $11.3 million per MW this year.
At that level, a 100 MW facility can cost around $1.1 billion before the tenant fills it with computing equipment.
The technology inside an AI facility can cost even more. JLL estimates that high-end tenant fit-outs can reach roughly $25 million per MW. A fully equipped 100 MW AI deployment can therefore involve several billion dollars of additional hardware.
Lenders already price the tenant risk differently. JLL says projects backed by non-investment-grade tenants can borrow at spreads roughly 200 to 300 basis points wider than projects backed by investment-grade hyperscalers.
The larger risk sits with the companies funding the computing itself. Amazon now plans around $220 billion of annual capital expenditure. Microsoft expects roughly $190 billion. Alphabet has raised its range to $195 billion-$205 billion, while Meta expects $130 billion-$145 billion.
That scale is beginning to affect cash generation. Alphabet's latest quarterly capital expenditure reached $44.9 billion, more than double the year-earlier level, and quarterly free cash flow turned negative at roughly $5.9 billion. Meta's free cash flow fell to less than $1 billion in its latest quarter as infrastructure spending climbed.
The spending still has substantial revenue growth behind it. AWS grew 37% in its latest quarter. Google Cloud grew 82%. Microsoft continues to say demand exceeds the compute capacity it can bring online.
Alphabet's cloud business is also becoming far more profitable. Google Cloud generated roughly $8.8 billion of operating profit in the latest quarter, with an operating margin around 35.6%.
The risk to watch is whether cloud and AI revenue eventually slow while infrastructure spending stays near today's extraordinary levels.
If you want more recent data on this point, please see our latest data center market report.
Could better AI chips make all these new data centers unnecessary?
Better chips will reduce the infrastructure needed for a given amount of AI work, but current usage growth is still outrunning those efficiency gains.
This is one of the hardest parts of any long-term data-center forecast because computing does not behave like a fixed demand pool.
Suppose a new accelerator cuts the cost of an inference by half. Total compute demand falls only if people keep doing roughly the same number of inferences. If lower prices cause AI to appear inside search, advertising, software development, customer service, video, autonomous systems and millions of business workflows, total computation can still increase dramatically.
The behavior of the hyperscalers currently supports that second scenario. Amazon continues to describe compute as capacity constrained despite constantly improving its own chips and infrastructure. Microsoft is bringing more GPUs, CPUs and storage online while expecting constraints to continue. Google is expanding TPU capacity at the same time that Google Cloud growth accelerates.
The workload mix is also changing. AI training favors very large centralized clusters. Inference can create demand closer to users because response time becomes more important when AI is embedded inside everyday applications.
Efficiency remains a serious source of forecast error, but so far better hardware has come with more total infrastructure demand rather than less.

This chart, featured in our data center market deck, shows how hyperscale AI-ready campus technology has evolved over time
Does electricity use prove the Data Center Market is really growing?
Yes, electricity consumption confirms that Data Center Market growth has already become large enough to change national power demand.
The International Energy Agency estimates that data centers accounted for around half of the increase in U.S. electricity consumption in 2025.
The IEA expects roughly the same share through 2030. U.S. electricity demand is projected to add more than 420 TWh over the next five years, with data-center expansion responsible for about half of that increase.
Globally, the IEA expects data-center electricity consumption to more than double to around 945 TWh by 2030. That would be slightly more electricity than Japan consumes today.
Lawrence Berkeley National Laboratory recently reached a similarly striking conclusion using a bottom-up model based on expected server shipments, equipment power use and cooling requirements. Its reference case puts U.S. data centers at 649 TWh in 2030, equivalent to 11.8% of national electricity consumption. The modeled range is wide, from 9.5% to 15.3%, which tells us how uncertain the exact endpoint remains.
These electricity numbers give us an independent check on the real-estate data: the expansion is already large enough to show up in national energy demand.
Will the Data Center Market keep growing through 2030?
The Data Center Market is very likely to keep growing through 2030, although today's most extreme growth rates will not survive everywhere.
Several separate pipelines point in the same direction.
Synergy expects hyperscalers to control 67% of worldwide data-center capacity by 2031, with hyperscale capacity growing more than threefold from today's level. It still expects ordinary colocation capacity to grow at close to double-digit annual rates.
JLL already tracks 66 GW under construction in North America. Because 95% is precommitted, a substantial part of the next several years of growth has customers attached to it before the buildings open.
The electricity forecasts also assume continued expansion. The IEA expects worldwide data-center power consumption to more than double by 2030. Lawrence Berkeley's U.S. reference case implies hundreds of additional terawatt-hours of consumption.
Still, we would not treat every announced gigawatt as future operating capacity.
Power connections can fail. Communities can reject projects. Financing can become expensive. AI companies can disappear. Chip efficiency can improve faster than expected. Some regions will inevitably build too much capacity.
The direction looks much more certain than the exact size.
If you want more recent data on this point, please see our latest data center market report.

In our data center market deck, we identify pain points entrepreneurs should prioritize
So, is the Data Center Market growing now?
Yes. The Data Center Market is growing very fast now, and the evidence is strong enough that we can say this with high confidence.
The clearest part is how many independent measurements agree.
Physical inventory is rising at double-digit rates across North America, Europe, Asia-Pacific and Latin America. In the major North American markets, users have been absorbing new capacity almost as quickly as it gets delivered. Vacancy remains around historic lows in the tightest hubs.
Customers are also reserving capacity years in advance. Much of the North American construction pipeline is already precommitted, while operators such as Digital Realty and Equinix continue to report stronger bookings and growing recurring businesses.
Big Tech has accelerated infrastructure spending again rather than pulling back. Amazon, Microsoft, Alphabet and Meta together are currently planning close to three quarters of a trillion dollars of annual capital expenditure, while AWS and Google Cloud are simultaneously posting some of their strongest growth in years.
The electricity system is showing the effect too. Data centers already account for around half of recent U.S. electricity-demand growth.
There are weak spots. Querétaro has seen vacancy rise after a huge supply increase. Some Chinese facilities have struggled with utilization. Older data centers cannot necessarily support modern high-density computing. Individual AI projects will fail, and some planned campuses will never receive enough power to open.
Our judgment is clear: the Data Center Market is in a major growth cycle right now. Demand is already visible in occupied megawatts, rents, operator bookings, cloud revenue, capital spending and electricity consumption.
OUR METHODOLOGY
This analysis tests whether the Data Center Market is growing now by looking at the parts of the market that should move if the growth is real: physical capacity, absorption, vacancy, rents, signed leases, operator revenue, hyperscaler infrastructure spending, construction commitments, power constraints, financing and electricity consumption.
We deliberately avoid relying on a single market-size estimate because data-center research firms often measure different things. Some count buildings and power infrastructure, while others add servers, networking, managed services or pieces of the cloud market. For this article, the cleaner unit is physical computing capacity and the economic activity attached to it.
Recent operating evidence gets more weight than long-range forecasts. Delivered megawatts, net absorption, vacancy, leases, bookings, current capital expenditure and electricity use tell us more about the market today than announced campuses that may not receive power or financing for years.
Where we use forward-looking evidence, we distinguish between stages. Capacity already operating is the strongest evidence. Capacity under construction and precommitted is also meaningful. Unleased construction gets more caution, and announced or planned campuses are treated as much less certain.
We also compare independent measures against one another. Capacity growth is checked against absorption and vacancy. Construction pipelines are checked against precommitments. Operator bookings are checked against revenue. Hyperscaler spending is checked against cloud growth and reported capacity constraints. Real-estate expansion is checked against electricity demand.
Geographic comparisons rely mainly on CBRE's Global Data Center Trends 2026 and North America Data Center Trends H2 2025, with JLL's North America Data Center Report Midyear 2026 used for the broader hyperscale construction, absorption, financing and frontier-market picture. Synergy Research Group is the main source for hyperscale facility counts, ownership mix and global capacity shares.
Commercial operator evidence comes directly from Digital Realty and Equinix. Their Q2 2026 results are used for revenue growth, bookings, signed lease backlog, renewal pricing, recurring revenue, interconnections and active expansion projects.
Hyperscaler demand and spending are grounded in company filings and earnings materials from Amazon, Microsoft, Alphabet and Meta. We use those sources for AWS and Google Cloud growth, 2026 capital-expenditure plans, management comments on capacity constraints and the split between servers, data centers and networking.
Electricity demand provides an independent validation layer. The International Energy Agency is the main source for global data-center electricity consumption and the share of U.S. demand growth linked to data centers. Lawrence Berkeley National Laboratory provides the bottom-up U.S. 2030 electricity-use scenarios, while the U.S. Energy Information Administration is used for the recent Texas load-forecast change.
China's capacity story is treated separately because utilization and policy coordination matter as much as raw construction. For that section, we use official Chinese sources around the National Integrated Computing Power Network and the East Data, West Computing initiative, alongside Synergy's hyperscale market data.
Key sources used for this analysis include: CBRE's Global Data Center Trends 2026, CBRE's North America Data Center Trends H2 2025, JLL's North America Data Center Report Midyear 2026, JLL on H1 2026 data-center demand, Synergy Research Group on hyperscale capacity, Digital Realty's Q2 2026 results, Equinix's Q2 2026 results, Amazon's Q2 2026 results, Microsoft's FY2026 Q3 earnings materials, Alphabet's Q2 2026 results, Meta's Q2 2026 results, the IEA's Energy and AI report, the IEA's Global Energy Review 2026, Lawrence Berkeley National Laboratory's United States Data Center Energy Usage Report: 2025 Update, the U.S. Energy Information Administration's Short-Term Energy Outlook, and official Digital China material on the national computing network.

This chart, featured in our data center market deck, shows the revenue mix by region across Europe, Asia, North America, Africa, and South America in the data center market
Who is the author of this content?
NEW MARKET PITCH TEAM
We track new markets so founders and investors can move fasterWe build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.