Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Microsoft did not abandon Project Natick because the underwater data center failed technically. The company confirmed in June 2024 that it was no longer building subsea data centers, even though its Scotland prototype ran for more than two years, used no cooling water and recorded fewer server failures than its land-based comparison. Natick demonstrated that sealed, remotely operated data centers can work. It did not demonstrate that they are the most economical or flexible way to run rapidly changing cloud and AI hardware.
What Project Natick was designed to do
Project Natick was Microsoft Research’s attempt to move some data-center infrastructure beneath the sea. The concept began in 2013 and treated the facility like a standardized submarine: build the module in a factory, ship it to a coastal site, lower it onto the seabed, connect power and fiber, and operate it without routine on-site staff.
The underwater setting addressed several conventional data-center problems at once:
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problems- Seawater could absorb heat without evaporating freshwater.
- A sealed, dry-nitrogen atmosphere could limit oxygen, humidity and some corrosion risks.
- Stable low temperatures could support efficient cooling and potentially longer hardware life.
- Coastal placement could put computing close to population centers and connected to offshore renewable energy.
- A factory-built module could, in Microsoft’s description, go from a deployment decision to power-on in under 90 days.
The operating model was “lights out,” not maintenance-free. Microsoft envisioned leaving the vessel submerged for up to five years, then recovering it to replace the server fleet, test and reseal the pressure vessel, and redeploy it.
#1 Best Overall
- Save valuable floor space: 6U wall mount server cabinet Dimensions: 13.78" H x21.65" W x17.72" D.Maximum mounting depth is 14.2"
- Keep critical network equipment secure: glass door and side panels are lockable to prevent unauthorized access. Front door can be installed on either side of the front of the cabinet to satisfy your door swing orientation preference
- Easy equipment configuration: Fully adjustable mounting rails and numbered U positions, with square holes for easy equipment mounting with top and bottom punch-out panels for easy cable access
- Durability: Made of high quality cold rolled steel holds up to 110lb (50kg) (Easy Assembly Required)
- PCI & HIPPA and EIA/ECA-310-E compliant
What Microsoft actually built
| Milestone | What happened |
|---|---|
| 2013 | Microsoft says the Natick concept originated. |
| 2015 | A first prototype spent 105 days in the Pacific Ocean. |
| 2018 | The full-scale Northern Isles module was deployed off Orkney, Scotland. |
| July 2020 | The vessel was retrieved after roughly two years underwater. |
| June 2024 | Microsoft executive Noelle Walsh told Data Center Dynamics: “I’m not building subsea data centers anywhere in the world.” |
The Scotland module sat about 117 feet (36 meters) below the surface. Microsoft’s project materials describe a vessel containing 855 servers and designed for remote operation. The company has retained the research and said it would apply lessons to other data-center technologies, including liquid immersion.
What the experiment proved
Reliability was better in the test comparison
Microsoft reported six failed servers among the 855 in the underwater vessel, versus eight failed servers among 135 land-based comparison servers. It characterized the underwater result as one-eighth the failure rate of its comparison population.
That is a notable result, but it is not a universal claim that every underwater data center is eight times more reliable. The populations were small and specialized, the test lasted roughly two years, and the comparison reflects Microsoft’s methodology. The result does suggest that a sealed environment with no routine human disturbance can reduce some hardware failures.
Free tools Windows power users keep installed
One-click scans. No signup required.
Cooling used no freshwater
Microsoft reported a power usage effectiveness (PUE) of 1.07 for the prototype and zero water consumption for cooling. The ocean supplied a large heat sink without the evaporative cooling systems used by many land facilities. That made Natick an impressive demonstration of low-water operation.
Remote operation was viable
The module operated without technicians working inside it. Microsoft monitored the equipment remotely and later recovered the vessel. The prototype endured pressure, currents, storms and marine growth; biofouling still had to be accounted for, but the vessel survived the test environment.
Deployment and recovery were possible
Microsoft’s recovery operation used a gantry barge, winches, robotics and marine specialists. Deployment and retrieval each took a full day in suitable conditions. That proved the procedure could be executed—not that it would be simple or cheap to repeat across a commercial fleet.
Why technical success did not make Natick a commercial product
Maintenance became a marine logistics problem
On land, a technician can replace a failed server, GPU, power supply, cable or storage device in a planned maintenance window. An underwater module cannot be opened at the seabed. A serious repair could require rerouting workloads, sending a marine crew, recovering the vessel, transporting it to shore, opening the pressure vessel, replacing equipment, testing and resealing it, then redeploying it.
Natick avoided routine visits; it did not eliminate maintenance. It exchanged frequent, relatively small interventions for infrequent, specialized and potentially high-consequence recovery operations.
Hardware refresh cycles moved faster than the concept
The planned multi-year submerged cycle is easier to imagine for a stable server fleet than for modern accelerated computing. Cloud operators increasingly need to change GPU generations, rack power levels, networking, memory and storage configurations, firmware and accelerator layouts.
Rank #2
- Save valuable floor space: 12U wall mount server cabinet Dimensions: 24.25" H x21.65" W x17.72" D. MAXIMUM MOUNTING DEPTH is 14.2".
- Keep critical network equipment secure: glass door and side panels are lockable to prevent unauthorized access; Front door can be installed on either side of the front of the cabinet to satisfy your door swing orientation preference
- Easy equipment configuration: Fully adjustable mounting rails and numbered U positions, with square holes for easy equipment mounting with top and bottom punchout panels for easy cable access
- Durability: Made of high quality cold rolled steel holds up to 110lb (50kg) (Easy Assembly Required)
- PCI & HIPPA and EIA/ECA-310-E compliant
A sealed module that is difficult to upgrade can leave useful capacity tied to yesterday’s hardware. This flexibility problem is central to the commercial critique of Natick. Industry commentary cited by ITPro identified the difficulty of upgrading, altering and servicing underwater systems as a limiting factor.
Marine deployment added a permanent infrastructure layer
A commercial installation would need suitable ports and vessels, subsea engineering, power and fiber cables, marine insurance, emergency-response capability and trained crews. Weather could constrain recovery windows. The Scotland test used specialized equipment and a carefully choreographed operation.
Recommended Free Tools
Microsoft also noted that marine space is regulated. A deployment needs an acceptable seabed, depth, currents and weather conditions, a power connection, fiber access, safe installation and recovery routes, and approval from authorities that control activity in the water.
The geography was narrower than the headline suggested
An underwater module only makes sense where the location provides both a technical and business advantage. It must be close enough to users or data sources to justify an edge-computing role, yet have suitable seabed conditions and connectivity. The ocean is not an empty, permit-free substitute for land.
Cooling was only one line in the total-cost calculation
The relevant question was never simply whether seawater could cool servers. It was whether the savings in cooling energy and freshwater outweighed the cost and risk of pressure vessels, marine installation, subsea cables, permits, recovery, insurance, spare capacity and lost hardware flexibility.
Microsoft’s project documentation said total cost of ownership—including manufacture, deployment, operation and recovery—was something the second phase was intended to study. Microsoft has not published a current public cost model showing that a commercial underwater fleet beats conventional facilities or modern liquid-cooled data centers at scale. The public record therefore does not support the simpler claim that Natick was “too expensive” as a confirmed internal postmortem.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Why AI made the fit harder
AI training and high-performance computing combine high rack power with unusually rapid hardware evolution. Operators may need new accelerators, denser memory, faster interconnects and different thermal designs before a five-year submerged cycle ends.
That makes access and incremental upgrades more valuable. This is an industry-context inference, not a publicly stated Microsoft reason for ending Natick. Microsoft’s announcement did not provide a detailed cost or workload postmortem. The inference follows from the mismatch between a sealed, recovery-based lifecycle and the fast-changing requirements of AI infrastructure.
Land-based liquid cooling solved more of the problem
Natick’s thermal insight did not require putting the entire data center underwater. Operators can now use liquid systems inside conventional facilities while retaining access to servers:
Rank #3
- Sturdy:4u server rack is construct from cold rolled steel, with a weight capacity of 110lbs(50kg); Electrostatic powder coat prevents rust and corrosion,quality finish
- Direct use:Open and use, not having to assemble it.Network rack can be placed flat or mounted on the wall,also can be installed vertically under the table
- Design Features:maximum mounting depth of 14 in,cables can be fixed on the side panel;Open frame server rack achieves effortless inspection, replacement and assemble
- Installation:wall mount network rack is easy to install,with instructions or videos for reference;Equipped with multiple accessories, suitable for different needs
- Application:EIA/ECA-310-E Compliant;wall mounted 4u rack fits all 19" racks and cabinets to hold various IT, network, and AV equipment;wall mount rack available in 4U, 6U, and 8U to choose
- Direct-to-chip cold plates: coolant removes heat directly from processors and accelerators.
- Rear-door heat exchangers: liquid systems retrofit many existing racks without changing every server.
- Coolant distribution units: these manage the facility loop for dense AI and HPC deployments.
- Single-phase and two-phase immersion: servers are placed in engineered dielectric fluids, with service procedures and hardware compatibility managed on land.
- Air/liquid hybrids: conventional cooling handles lower-density equipment while liquid systems handle the hottest racks.
For example, Vertiv lists CoolChip CDU configurations from approximately 70 kW to 2,300 kW, depending on configuration, while Schneider Electric and Motivair market liquid-cooling infrastructure for accelerated-compute environments. Both are quote-based offerings rather than products with public list prices. These approaches preserve technician access and allow staged upgrades—advantages a submerged pressure vessel does not provide.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWhat Microsoft said about the project’s end
Microsoft’s public language was not that Natick failed. Walsh said “it worked.” The company said it learned about operations below sea level, vibration and effects on servers, and would apply those lessons to reliability, sustainability and liquid-immersion research.
The accurate description is therefore: Microsoft ended subsea data centers as an active buildout strategy while retaining Natick’s research. That is a commercial decision, not a declaration that the engineering experiment was useless.
Could underwater data centers still return?
Possibly, but the likely use cases are narrower than general-purpose hyperscale cloud capacity. The concept could fit:
- Remote coastal edge computing with stable workloads.
- Military or tactical deployments where physical remoteness is useful.
- Locations with scarce freshwater but reliable marine access.
- Offshore-energy or maritime applications.
- Workloads that can run for years without hardware changes and have redundant capacity elsewhere.
These are potential use cases, not evidence that Microsoft currently plans to commercialize Natick. Any operator would still need to assess cable damage, pressure-vessel leaks, corrosion, biofouling, storms, permitting, environmental objections, vessel-level redundancy and the possibility that hardware becomes obsolete before recovery.
What operators should choose instead
| Need | More practical direction | Main trade-off |
|---|---|---|
| High-density AI or HPC cooling | Direct-to-chip liquid cooling and coolant distribution units | Requires plumbing, controls and facility changes |
| Existing-rack retrofit | Rear-door heat exchangers | Capacity and rack compatibility vary |
| Very high thermal density | Single-phase or two-phase immersion | Fluid compatibility and service procedures become critical |
| Compute close to users or sensors | Edge appliances such as Azure Stack Edge | Not a substitute for a large hyperscale GPU cluster |
| Fast deployment where grid capacity is constrained | Modular power-and-cooling packages such as Vertiv BYOP&C | Enterprise-scale solution with quote-based pricing |
Azure Stack Edge addresses the proximity problem without subsea deployment; Microsoft’s pricing page says displayed estimates are not quotations and that shipping and other charges may apply. Modular power-and-cooling systems address deployment speed while keeping equipment accessible on land.
The bottom line
Project Natick was a successful engineering experiment that failed to become Microsoft’s preferred commercial architecture. It showed that a sealed underwater module could run reliably, achieve a reported 1.07 PUE and use no cooling water. It also exposed the cost of giving up everyday access to hardware: recovery logistics, permits, specialized marine infrastructure and a deployment cycle that is poorly matched to rapidly changing AI systems.
The ocean solved cooling and some reliability problems. It did not solve the broader business problem of delivering adaptable compute at the lowest lifecycle cost. For most operators, land-based direct-to-chip, rear-door or immersion cooling offers much of Natick’s thermal benefit without surrendering upgradeability.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
