Modular data centers: which startup is ahead?
In our data center market deck, you will find everything you need to understand the market
SUMMARY
Crusoe is ahead in modular data centers today, with Armada emerging as the only startup close enough to challenge its overall lead.
The real separator is not funding, announced capacity or one impressive technical number. It is the combination of operating infrastructure, repeat customers, manufacturing readiness, energy access and a way to monetize the computing capacity.
Crusoe has the strongest proof of repeat demand. Redwood Materials operated four Spark units before agreeing to expand the same installation to 24, which carries more weight than a large first-time announcement.
Armada has assembled the more unusual customer list. Its systems have reached oil and gas operations, the U.S. Navy, a Navy warship and an offshore drilling installation, giving it a strong position in remote and sovereign computing.
The main weakness in Armada’s case is disclosure. Its reported bookings growth is extremely fast, but the company does not publish the starting values, operating megawatts or total number of deployed units.
Nautilus and ECL show why specialist leadership is different from overall leadership. Nautilus has the deeper liquid-cooling operating record, while ECL reports the lowest PUE from a live modular AI facility.
Manufacturing may decide the next phase of the race. Crusoe has committed more than $200 million to a 352,000-square-foot factory and its first manufactured fleet, while Armada and Johnson Controls are preparing a facility of up to 400,000 square feet.
Speed is the strongest economic argument for modular infrastructure. Bringing scarce GPUs online in a few months can be more valuable than achieving a slightly lower construction cost after waiting years for a conventional campus and grid connection.
Submer has made the biggest strategic leap lately. It has moved from immersion cooling toward cloud infrastructure and full data center development, although most of its announced gigawatt-scale portfolio remains a pipeline rather than operating capacity.
ECL has produced the most striking result relative to its disclosed venture funding. With about $17 million raised, it opened a fully leased 1 MW facility running dense production AI hardware, giving it more physical proof than several better-funded competitors.
The current ranking is Crusoe first, Armada second, Nautilus third, Submer fourth, ECL fifth, Flexnode sixth and Metrobloks seventh. Armada could narrow the gap quickly, but it now needs to show Leviathans leaving its factory and operating at meaningful scale.
This market map, featured in our data center market deck, highlights top companies and startups in the data center market
Which startups actually belong in the modular data center race?
Seven startups deserve a place in this modular data center comparison, although only Crusoe and Armada currently look like broad category leaders.
We include companies that are building complete modular data centers, operating modular facilities, or supplying a large enough share of the infrastructure to compete for the same AI projects. That gives us Crusoe, Armada, Nautilus Data Technologies, Submer, ECL, Flexnode and Metrobloks.
The comparison still needs some boundaries. Crusoe and Armada sell integrated computing infrastructure. Nautilus and Submer built their positions around cooling before moving closer to complete data center delivery. ECL combines modular buildings with its own energy architecture. Flexnode manufactures prefabricated AI data halls, while Metrobloks develops permanent facilities using repeatable modular designs.
We leave out Vertiv, Schneider Electric, Eaton, ABB, Huawei and Rittal because they are established industrial groups. They remain important competitors. Vertiv already sells prefabricated systems ranging from small modules to 50 MW campuses, while Schneider Electric can deliver complete data center pods through a global manufacturing network.
We also exclude EPG from the startup ranking. The company was founded in 2004, operates large manufacturing hubs in Asia and has supplied customers including GDS and Huawei. It works better as a private industrial benchmark for the startups below.
Funding figures require some care. Crusoe’s total supports cloud computing, energy projects and hyperscale campuses alongside its Spark modular business. Submer’s total varies between databases because some sources count convertible instruments and debt differently. Flexnode has raised at least $38.9 million across its disclosed seed and Series A rounds.
| Startup | What it currently does | Funding raised |
|---|---|---|
| Crusoe | Builds and operates AI infrastructure, including Spark modular AI factories | About $3.9B across debt and equity for the wider company |
| Armada | Builds rugged Galleon units and megawatt-scale Leviathan modular data centers | Nearly $500M |
| Nautilus Data Technologies | Supplies prefabricated liquid-cooling infrastructure and operates a liquid-cooled facility | About $223M |
| Submer | Builds immersion-cooling systems and is expanding into cloud and full data center development | About $102M, with some databases reporting lower totals |
| Flexnode | Builds prefabricated high-density AI data halls with integrated power and cooling | At least $38.9M |
| Metrobloks | Develops modular, liquid-cooled data centers in urban and regional markets | About $20.8M |
| ECL | Builds modular AI data centers powered through grid, gas, batteries and hydrogen fuel cells | $17M |
Is one modular data center startup clearly ahead today?
Crusoe is ahead in modular data centers today, with Armada close enough to make the race genuinely competitive.
Crusoe has more operating history, more capital and a much larger infrastructure business behind its modular product. The company says it has deployed more than 400 earlier modular units in difficult energy locations. Its newer Spark system also has a live customer installation that is moving beyond its initial trial.
Armada has built a more focused modular data center company. Its Galleon units have been used in Saudi industrial operations, delivered to the U.S. Navy, tested aboard a Navy warship and selected for an offshore Aker BP deployment. Its newer Leviathan product brings the same idea into megawatt-scale AI computing.
The gap comes from proof of scale. Crusoe has disclosed operating units, a repeat customer expansion and more than $200 million committed to a dedicated factory and its first manufactured fleet. Armada has strong customers and a large planned factory, although it still keeps its installed unit count and operating megawatts private.
Nautilus, Submer and ECL lead narrower parts of the market. Nautilus has the longest liquid-cooling operating record. Submer has the strongest immersion-cooling ecosystem. ECL has the clearest live example of a hydrogen-powered modular AI facility.
Flexnode and Metrobloks remain further back. Both have credible partners and ambitious designs, yet neither has publicly identified a commissioned commercial facility built around its current platform.
If you want more recent data on this point, please see our latest data center market report.
As this chart shows, and as featured in our data center market deck, search interest in data centers has increased significantly
Who has actually put modular data centers into the field?
Crusoe has deployed the most modular infrastructure, while Nautilus, ECL and Armada provide the clearest evidence that their systems can handle real customer workloads.
Crusoe’s figure of more than 400 deployed modules gives it the deepest field experience, although most of those units came before Spark and were used for energy-site computing. We should view the number as evidence of operational experience, rather than as 400 identical modern AI factories.
The more useful Spark example sits at Redwood Materials in Nevada. Four units began operating beside a solar and second-life battery microgrid in 2025. Crusoe and Redwood later agreed to expand the site to 24 units and nearly seven times its original computing capacity.
Nautilus has a smaller footprint with a longer facility-level record. Its Stockton site provides 6.5 MW across four computing vaults, reports a PUE of 1.15 and has accumulated more than 500,000 operating unit-hours. Nautilus has also delivered 12 MW of cooling capacity to Start Campus in Portugal.
ECL’s first commercial site is only about 1 MW, although it is fully leased and has hosted Lambda’s production-grade NVIDIA systems. Armada has shown that its smaller Galleons work in industrial, military and maritime conditions. The company has yet to publish a comparable total for live units or megawatts.
| Startup | Strongest operating evidence | What remains unclear |
|---|---|---|
| Crusoe | 400+ earlier modular units and a live Spark installation expanding from four units to 24 | How many earlier units are technically comparable with Spark |
| Armada | Deployments involving Aramco Digital, Microsoft, the U.S. Navy and offshore energy operations | Total operating units, installed MW and live Leviathan capacity |
| Nautilus | A 6.5 MW facility, 500,000+ operating unit-hours and a 12 MW cooling delivery | Repeat sales of complete modular facilities beyond cooling systems |
| ECL | A fully leased 1 MW modular facility running production AI equipment | Replication across several sites and much larger capacities |
| Submer | Commercial immersion-cooling installations and a delivered 100 kW MegaPod | Operation of the complete campuses it is now announcing |
| Flexnode | No named commissioned customer facility found | Whether the Eaton-backed product performs as planned in live use |
| Metrobloks | Several development sites and partnerships | The company’s first completed and operating data center |
Who has turned modular data centers into real customer demand?
Crusoe currently has the strongest evidence of repeat demand, while Armada has assembled the most impressive mix of strategic customers.
Crusoe’s work with Redwood is valuable because the customer used the first four units before committing to another 20. That sequence tells us more than a large initial announcement. The installation is being expanded after real operation, using the same energy platform and location.
Crusoe also signed a framework with Energy Vault covering phased Spark deployments of up to 25 MW. We treat that figure cautiously because a framework can change before the entire capacity is ordered and commissioned.
Armada’s customer list is broader. Aramco Digital and Microsoft used its systems in Saudi Arabia. The U.S. Navy received a Galleon and later tested one during the UNITAS maritime exercise. Aker BP has agreed to deploy another unit on an offshore drilling installation.
These customers operate in environments where conventional data centers are impractical. That supports Armada’s product positioning. The commercial depth remains harder to judge because Armada does not disclose the number of units per customer, contract values or repeat-order volumes.
Nautilus provides cleaner capacity figures. Its Stockton facility was reported as 86% leased when it was offered for sale, and Cerebras had signed for 2.5 MW. Start Campus then deployed 12 cooling distribution units representing 12 MW of capacity.
ECL has the strongest demand evidence relative to its size. Its 1 MW MV1 site reached full occupancy, while Lambda later doubled its footprint there. A company with only $17 million in disclosed funding has already persuaded customers to run dense production hardware on its infrastructure.
If you want more recent data on this point, please see our latest data center market report.
This chart, featured in our data center market deck, illustrates yearly venture capital funding for data center startups
Which modular data center startup is growing fastest now?
Armada reports the fastest commercial growth right now, while Crusoe shows the strongest physical expansion that we can independently compare.
Armada said customer bookings grew 540% between fiscal 2025 and fiscal 2026. It then reported a 2,000% year-over-year increase during the first quarter of fiscal 2027. Those percentages show rapid acceleration, although Armada has not published the dollar value behind them.
A company moving from $1 million to $21 million would also report 2,000% growth, so the starting point matters. Armada’s new $230 million financing and planned factory suggest that the underlying demand is substantial. The percentages still do not tell us how much revenue or installed capacity the bookings will ultimately produce.
Crusoe’s clearest growth measure is the sixfold expansion in Spark unit count at Redwood. Computing capacity is expected to rise almost sevenfold because the newer units are denser. This looks slower than Armada’s percentages, but it gives us a physical before-and-after comparison.
Submer is expanding in a different way. It acquired Radian Arc, added a cloud layer through InferX and launched Rubix Data Centers with an initial portfolio of more than 8 GW of powered land. Submer has therefore moved from a cooling specialist toward a full AI infrastructure group in less than a year.
That shift could produce the largest long-term change in the ranking. Most Rubix capacity still belongs to the development pipeline, so we would not place Submer above the operating leaders yet.
Whose modular data center product is ready to buy and deploy?
Crusoe has the most mature complete AI product, while Armada is currently the strongest option for smaller systems that must work in remote or difficult locations.
Crusoe says a Spark installation can become operational in roughly three months. Each unit arrives with power distribution, cooling, fire protection, networking and GPU-ready racks already integrated. The Redwood deployment shows that the design can operate beside a non-traditional power system.
Armada says a Galleon can be deployed in about 60 days. Its public deployments support that claim for smaller edge systems. Galleons have operated around oil and gas infrastructure, military networks and maritime equipment where building a permanent data center would take far longer.
Armada’s megawatt-scale Leviathan product is less proven. The company introduced Leviathan in 2025 and planned continuous factory production for summer 2026. Its latest public materials still provide no production total or named operating Leviathan site.
Nautilus offers a mature cooling product rather than the most complete computing platform. It quotes delivery periods of roughly 12 to 16 weeks for some cooling distribution units, compared with the much longer waits affecting conventional infrastructure components.
Flexnode says its prefabricated process cuts deployment schedules by an average of 35%. Eaton now supplies power backup, racks, cabling and 800-volt direct-current infrastructure for the NX Compute Module. That partnership makes the product more credible, although Flexnode has not published the customer projects behind its 35% figure.
ECL has proven that its basic design works at 1 MW. The next test is much harder: repeating the system across a planned 35 MW campus while coordinating utility power, gas generation, batteries and hydrogen fuel cells.
This chart, featured in our data center market deck, shows how Equinix is capturing share in data centers
Who can manufacture modular data centers at real scale?
Crusoe has made the largest concrete manufacturing commitment, while Armada is building the factory most likely to challenge it.
Crusoe’s Spark Factory covers 352,000 square feet in Colorado. The company has committed more than $200 million to the facility and its first fleet, with the first factory-produced units expected during the third quarter of 2026.
That gives Crusoe a credible route to repeated production. It has already designed, installed and operated hundreds of earlier modules, so the company is industrializing a process it knows instead of starting from a blank sheet.
Armada and Johnson Controls plan to manufacture modular data centers from a facility covering up to 400,000 square feet in Arizona. The factory is expected to create more than 500 jobs and begin with the Leviathan product.
The physical site would be slightly larger than Crusoe’s factory. We still need output data. Armada has not disclosed how many Leviathans it expects to manufacture each year, and there is no public confirmation of sustained production volumes.
Flexnode follows a more asset-light path. Eaton can supply critical electrical equipment and manufacturing support, allowing Flexnode to sell data halls from 3.5 MW to 35 MW without building every component itself. This may become an efficient model once a customer facility is running.
Nautilus also relies on manufacturing and integration partners. That approach already supports shorter delivery periods for cooling equipment. Building an entire AI data center requires a wider group of suppliers and a much larger project team.
Metrobloks sits slightly outside this factory race. Its proposed Liberty campus would provide 150 MW across three permanent buildings, while other sites are planned in Miami, Indianapolis and Phoenix. Metrobloks is scaling a development pipeline, rather than shipping standardized data centers from a production line.
Who has the best cooling and energy efficiency?
ECL currently reports the lowest PUE in an operating modular AI data center, while Nautilus has the more convincing long-term operating record.
ECL reports a PUE of approximately 1.1 at its MV1 facility. Lambda has used the site for NVIDIA GB300 NVL72 systems at reported rack densities of about 142 kW, so the figure comes from a facility handling genuinely dense AI equipment.
Nautilus reports a PUE of 1.15 at Stockton. It also has more than 500,000 unit-hours of operating history and uses no consumptive water for cooling. That combination gives Nautilus a stronger reliability case, even though its published PUE is slightly higher.
The difference from a conventional new facility can be meaningful. At 1 MW of computing load, a PUE of 1.3 requires about 300 kW for cooling, power conversion and other overhead. A PUE of 1.15 reduces that overhead to 150 kW. ECL’s reported 1.1 would bring it down to roughly 100 kW.
These figures come from different facilities, climates and power systems, so this is not a controlled laboratory test. ECL has the better published number. Nautilus has much more operating evidence behind its claim.
Submer remains the strongest specialist in immersion cooling. Its technology has been tested in a certified combination involving Intel, Shell and Supermicro, while products have been deployed with companies including Castrol and Macquarie Data Centres. Some Submer product materials cite PUE as low as 1.03, although that remains a product-level claim rather than an audited campus result.
If you want more recent data on this point, please see our latest data center market report.
This chart, featured in our data center market deck, illustrates yearly funding for data center startups
Which modular data center startup offers customers the best economics?
Crusoe currently has the strongest economic case for urgent AI capacity, even though no startup publishes enough standardized pricing for a clean cost-per-megawatt comparison.
Crusoe’s main advantage is speed to revenue. A customer can place Spark units beside an available energy source and begin using expensive GPUs without waiting several years for a traditional campus and full grid connection.
That timing can outweigh a modest difference in construction cost. A delayed GPU cluster produces no revenue while the hardware depreciates. Crusoe also controls the energy relationship, data center infrastructure and cloud service, which can reduce the number of companies involved in getting a site online.
Nautilus has the clearest operating-cost argument. Its reported PUE of 1.15 implies about half the facility overhead of a site operating at 1.3 PUE. Zero-water cooling may also lower costs or remove a development obstacle in areas where water use is restricted.
ECL could save customers from long grid-interconnection waits. Its FlexGrid design combines several energy sources on the same site. The economics of hydrogen remain difficult to judge because ECL does not disclose fuel, storage, conversion and maintenance costs.
A low PUE only measures how efficiently a facility uses the electricity it receives. It says little about how much that electricity costs to generate. ECL must prove that hydrogen improves the complete customer bill, especially as its projects move from one megawatt to tens of megawatts.
Armada’s strongest economic use case appears in remote locations. Its customer may avoid sending large amounts of raw operational data over unreliable networks, while also avoiding the cost of constructing a permanent local facility. The value depends heavily on the workload and location, which makes a universal price comparison unrealistic.
What can the modular data center leaders do that competitors cannot easily copy?
Crusoe has the hardest overall system to reproduce, while Armada has built the strongest specialist position around remote and sovereign computing.
Vertiv, Schneider Electric and other industrial groups already know how to manufacture prefabricated power, cooling and computing infrastructure. A startup needs an advantage beyond the physical enclosure.
Crusoe combines energy development, modular manufacturing, large data center construction, cloud services and workload operation. It can locate power, build the computing environment and sell access to that capacity through Crusoe Cloud.
A competitor could create a similar module. Reproducing the energy relationships, operating experience, factory, cloud demand and customer contracts together would take much longer. Crusoe’s wider business strengthens Spark rather than distracting from it.
Armada combines rugged hardware with its Armada Edge Platform, satellite connectivity and centralized management. Its Azure Local integration also gives customers a familiar software environment inside locations where ordinary cloud access may be unreliable or restricted.
Experience in military, offshore and industrial deployments adds another layer. Large equipment manufacturers can build a strong box, although they may lack Armada’s software and knowledge of these operating environments.
Nautilus has patents and years of operating data around facility-scale liquid cooling. Submer has immersion-cooling expertise, certified hardware combinations and a growing customer ecosystem. Their technical positions are valuable, although the wider cooling market includes several well-funded incumbents.
ECL’s advantage comes from the way it combines energy sources and manages them through one platform. Some of the core fuel-cell hardware comes from partners such as PowerCell and Bosch, so ECL’s durable value will depend on project execution and energy orchestration.
If you want more recent data on this point, please see our latest data center market report.
This chart, featured in our data center market deck, compares the main business model options for hyperscale data center operators
Is the best-funded modular data center startup producing the best results?
Crusoe has converted the most capital into operating scale, while ECL has produced the clearest physical result from a small venture budget.
Crusoe has raised about $3.9 billion across debt and equity, nearly eight times Armada’s total. Armada has raised around 12 times more than Flexnode and almost 30 times more than ECL.
The totals are not directly comparable because the money supports very different businesses. Crusoe uses capital for cloud hardware, power projects and hyperscale campuses alongside Spark. Nautilus and Submer are more focused on cooling. Metrobloks finances site development, while ECL must eventually rely on project-level capital for large campuses.
Crusoe can still point to the most output. It has hundreds of earlier modular units, a repeat Spark customer, a funded factory and a large AI infrastructure business capable of filling new capacity.
Armada has turned nearly half a billion dollars into credible deployments, a full software platform, two modular product families and a major manufacturing partnership. Its efficiency remains hard to measure without revenue, unit volume or installed-capacity figures.
ECL has done something unusual with $17 million. It opened a fully leased 1 MW facility that runs production AI hardware. That result does not prove that ECL can finance and deliver a 35 MW campus, although it gives the company more operating credibility than its funding total would suggest.
Flexnode and Metrobloks have raised more than ECL while disclosing less live infrastructure. Their partnerships and pipelines may eventually justify the spending. Today, ECL has more physical proof per disclosed venture dollar.
Which modular data center startup has the most momentum lately?
Crusoe has the strongest momentum connected to operating infrastructure, while Submer has made the biggest recent strategic move.
Crusoe has recently opened its Spark manufacturing strategy, launched Edge Zones and expanded its wider contracted AI infrastructure capacity toward 5 GW. Only a small part of that 5 GW belongs to modular Spark systems, so we do not count the full figure as modular capacity. It still shows that Crusoe has customers and energy projects capable of supporting a much larger infrastructure platform.
Armada raised $230 million at a $2 billion pre-money valuation, announced its Johnson Controls factory and added the Aker BP offshore agreement. The next milestone is straightforward: the company needs to show Leviathans leaving the production line and operating for customers.
Submer has changed shape quickly. It acquired Radian Arc, launched the Rubix development platform and disclosed an initial portfolio of more than 8 GW of powered land. It also announced plans to invest more than €1 billion in an AI data center at a former industrial site in Flix, Spain.
These moves give Submer a much broader story than immersion cooling. Construction and customer commitments will determine whether this becomes a real operating advantage.
ECL’s latest progress is more concrete. PowerCell disclosed a firm order for hydrogen fuel-cell systems in the 5 MW class for ECL’s 35 MW CSC-1 campus. The companies also signed a separate memorandum covering about 300 MW of possible future capacity.
Metrobloks has added a proposed 150 MW campus in Missouri, a development joint venture with Lincoln Property Company and a potential 100 MW project with Soluna. Its pipeline has grown quickly, while completed capacity remains at zero.
This chart, featured in our data center market deck, shows the revenue mix across customer segments in the data center market
Can we trust the modular data center numbers?
Operating megawatts, repeat deployments and firm equipment orders deserve the most weight; bookings growth and planned gigawatts tell us much less on their own.
The strongest evidence involves equipment that can be located and customers that can be named. Nautilus’ Stockton facility, ECL’s MV1 site, Armada’s Navy delivery and Crusoe’s operating Spark units all pass that test.
A repeat order carries even more weight because the customer has already seen the product work. Crusoe’s Redwood expansion therefore influences our ranking more than a similarly sized first-time announcement would.
Firm purchase orders sit one level below operating infrastructure. ECL’s order for roughly 5 MW of PowerCell equipment has a stated value and delivery schedule. Its separate 300 MW memorandum describes a possible future relationship and deserves a much larger discount.
Development pipelines can be real assets, especially when the company controls land and power. They can also take years to finance, permit, pre-lease and construct. Submer’s 8 GW Rubix portfolio and Metrobloks’ 150 MW Liberty project show ambition and access to sites, rather than commissioned capacity.
Armada’s growth percentages have a different problem. A 540% or 2,000% increase sounds enormous, although the company has not disclosed the original booking values. We use the figures to measure direction and pace, without trying to convert them into market share.
Company-reported performance numbers also require context. Crusoe’s 400-plus modules include systems built before Spark. Submer’s very low PUE claims relate to specific cooling configurations. Flexnode’s 35% schedule reduction lacks a published project sample.
The evidence is uneven, so operating sites, repeat customers and disclosed capacity sit at the top of our ranking.
Which modular data center startups are actually ahead?
Crusoe is the startup ahead overall in modular data centers, Armada is the closest challenger, and the rest currently lead narrower pieces of the market.
Crusoe wins because it combines operating experience, repeat customer demand, dedicated manufacturing, access to energy and a cloud platform that can monetize the computing capacity. No other startup in the comparison has assembled all five at a similar scale.
Armada has the best chance of changing the ranking. It has raised enough money to compete, attracted serious industrial and government customers and partnered with Johnson Controls on a large factory. Confirmed Leviathan production and disclosed operating megawatts would narrow the gap quickly.
Nautilus takes third place because its technology has accumulated years of operating evidence. Its shift toward cooling infrastructure makes it a more specialized company than Crusoe or Armada, although that specialization also gives its efficiency claims more credibility.
Submer ranks fourth after a year of aggressive expansion. Its cooling technology is established, and Rubix gives it access to a much larger part of the value chain. The company now needs to turn powered land, acquisitions and announced investment into running campuses.
ECL ranks fifth with a small but real operation. Its first facility, Lambda deployment and firm fuel-cell order give it more credibility than many larger announced projects. The 35 MW CSC-1 campus will show whether ECL can repeat the model at commercial scale.
Flexnode ranks above Metrobloks because its product is clearly standardized and supported by Eaton’s industrial supply chain. Metrobloks has the larger development pipeline, although it currently looks more like a modular data center developer than a repeatable manufacturing company.
| Rank | Startup | Why it ranks here |
|---|---|---|
| 1 | Crusoe | The broadest combination of operating experience, customer expansion, manufacturing investment, energy access and cloud monetization |
| 2 | Armada | The strongest focused challenger, with high-quality deployments, rapid reported growth, strong software and a major factory partnership |
| 3 | Nautilus Data Technologies | The best long-term operating evidence among the specialists, including a live multi-megawatt facility and more than 500,000 unit-hours |
| 4 | Submer | The strongest immersion-cooling ecosystem and a rapidly expanding full-stack strategy, with most complete-campus capacity still ahead |
| 5 | ECL | A small but genuinely operating competitor with dense AI workloads, full occupancy and a firm order supporting its next 35 MW project |
| 6 | Flexnode | A credible modular data hall design backed by Eaton, currently held back by the absence of a named operating facility |
| 7 | Metrobloks | A fast-growing development pipeline with substantial planned capacity, although the company has yet to commission its first facility |
If you want more recent data on this point, please see our latest data center market report.
This chart, featured in our data center market deck, shows how hyperscale AI-ready campus technology has evolved over time
OUR METHODOLOGY
This analysis tests which modular data center startup is ahead based on what the companies have actually demonstrated. We compare the contenders across operating deployments, customer demand, commercial growth, product readiness, manufacturing capacity, cooling and energy performance, customer economics, defensibility and recent momentum.
We include startups that build complete modular data centers, operate modular facilities, or supply enough of the infrastructure to compete for the same AI projects. We exclude established industrial groups such as Vertiv, Schneider Electric, Eaton, ABB, Huawei and Rittal from the startup ranking, even though they remain important competitors.
We placed the greatest weight on physical and commercial execution: commissioned infrastructure, named customers, repeat deployments, operating capacity and firm equipment orders. A customer expansion after an initial deployment carries more weight than a first-time framework or a large capacity announcement.
Bookings growth, planned gigawatts, memoranda and development pipelines were used as indicators of direction rather than operating scale. We did not convert undisclosed bookings percentages into revenue or market share, and we did not count planned capacity as commissioned infrastructure.
Company performance claims were interpreted according to what they measure. Crusoe’s 400-plus modules show operating experience, but most predate Spark. Product-level PUE claims were separated from results reported by complete operating facilities, while planned factory size was kept separate from proven annual output.
Funding was used as context rather than as a ranking metric. Crusoe’s capital supports cloud infrastructure, energy projects and hyperscale campuses alongside Spark, while other companies use their funding for cooling equipment, modular manufacturing or site development.
We also separated specialist leadership from overall leadership. Nautilus, Submer and ECL can lead in liquid cooling, immersion cooling or alternative energy architecture without matching Crusoe and Armada across the entire modular data center value chain.
Key sources used for this analysis include Crusoe and Redwood Materials on the expansion from four to 24 Spark units, Crusoe on the original Redwood deployment, Crusoe on its Spark manufacturing facility, Armada on its financing, bookings growth and Johnson Controls factory, Armada and Aker BP on the offshore Galleon deployment, and Armada’s Galleon and Leviathan product information.
Additional operating and technical sources include Nautilus Data Technologies on its Stockton facility, ECL on its FlexGrid architecture and reported efficiency, PowerCell on its firm fuel-cell order and separate future-capacity memorandum with ECL, Submer on its acquisition of Radian Arc, Submer and Radian Arc on their Vietnam GPU deployment, and Metrobloks on its proposed Liberty campus.
In our data center market deck, we identify pain points entrepreneurs should prioritize
Related blog posts
- Who is leading the race in data center cooling?
- Which startup is redefining how data centers are powered?
- Which startup is setting the pace in AI data centers?
