NanoDC

NanoDC (Nano Data Center) is a factory-built data center developed by Liquid Energy for AI inference. The company describes a self-contained module combining computing equipment, solar generation, batteries, cooling and networking. Its November 2026 launch presented the system as a way to add inference capacity without waiting for a grid connection.^[1][2]

Design

Liquid Energy's announcement describes the following components. These are the company's published design claims, rather than independently measured specifications.^[1]

The company's website describes selling inference capacity to customers. It presents capacity growth as adding complete modules rather than extending a conventional building.^[2]

Deployment claims and availability

The launch announcement and the company website give different timing measures:^[1][2]

These intervals describe different stages. They are not published results from an independent deployment study.^[1][2]

In its Y Combinator profile, Liquid Energy said it had 20 units and manufacturing capacity of 2 MW per month, with a plan to reach 20 MW per month over the following year. The profile does not establish that all 20 units were serving customers. Nor does a manufacturing-rate claim specify the computing or electrical capacity of each unit.^[3]

The November announcement said the first units would become operational that month and that shipments of thousands of units would begin in 2027. Those statements were a delivery roadmap, rather than a report of thousands of completed deployments.^[1]

Developer

Y Combinator lists Liquid Energy as a Waterloo, Canada company founded in 2024 and a participant in its Fall 2026 batch. Its founder directory names Sid Ravikumar, Sajan Khosa and Arsh Sekhon. The company's website identifies Ravikumar as chief executive officer, Khosa as chief commercial officer and Sekhon as president.^[2][3]

Evidence and technical limits

The public launch does not provide an accelerator model, solar-array rating, battery capacity, measured inference throughput or uptime benchmark. The website's comparison of a conventional data center with a NanoDC fleet is labeled an illustrative model; it is not a live operational-results dashboard.^[1][2]

Research on renewable-powered computing helps explain the design problem but does not validate NanoDC. In a 2023 preprint, Seyed Morteza Nabavinejad and Tian Guo profiled inference workloads on an RTX 3090 and used renewable-power traces to study how batching and multiple model instances affect throughput under changing power limits. The experiments involved convolutional neural networks, rather than a NanoDC installation or a large language model serving fleet.^[4]

A separate study by Wedan Emmanuel Gnibga, Anne Blavette and Anne-Cecile Orgerie compared renewable-powered data-center configurations using grid connections, batteries, and batteries with hydrogen storage. It examined the infrastructure and cost tradeoffs involved in electrical autonomy.^[5] Neither paper measures NanoDC's cost, reliability or environmental impact. A solar-and-battery design alone therefore does not establish a particular cost per token or a measured lifecycle-emissions result for this product.

Continue reading

• AAI DatacenterAn AI datacenter (also written as AI data center) is a facility designed to run artificial intelligence workloads on dense groups of GPUs or other accelerators.Listed in See also
• AAI energy consumptionAI energy consumption is the electricity, and the associated water, land, and emissions, required to train and operate artificial intelligence systems, principally generative AI and large language models…Listed in See also
• IInference EconomicsInference economics is the study of the resources, costs and revenue associated with running a trained AI model. It connects the price of compute with the work a serving system delivers.Listed in See also
• IInferenceIn machine learning, inference is the execution of a trained model to produce an output from an input.Mentioned 6 times above
• DData CenterA data center is a purpose-built facility, or a dedicated part of a facility, that houses and interconnects information technology and telecommunications equipment together with the power…Mentioned 3 times above
• HHyperscalerA hyperscaler is a company that builds and operates computing infrastructure at a scale far beyond a conventional enterprise IT estate: globally distributed fleets of data centers holding millions of servers…Also in AI Energy

More to read

• xAI Colossus 2
• NVIDIA DSX MaxLPS
• AI data center moratorium wave (2026)
• Constellation Energy
• GE Vernova
• Oklo

Topics:AI EnergyAI InferenceAI InfrastructureData Centers

References

1. ^1 ^2 ^3 ^4 ^5 ^6Liquid Energy. NanoDC launch announcement. November 2026. Company description and deployment roadmap, reviewed November 7, 2026.
2. ^1 ^2 ^3 ^4 ^5 ^6Liquid Energy. AI inference powered by sunlight. Product and company website, reviewed November 7, 2026.
3. ^1 ^2Y Combinator. Liquid Energy: Self contained solar-powered AI inference modules: the NanoDC. Company profile and company-authored launch description, reviewed November 7, 2026.
4. ^Nabavinejad, Seyed Morteza, and Tian Guo. Opportunities of Renewable Energy Powered DNN Inference. arXiv:2306.12247, June 21, 2023. Preprint.
5. ^Gnibga, Wedan Emmanuel, Anne Blavette, and Anne-Cecile Orgerie. Renewable Energy in Data Centers: The Dilemma of Electrical Grid Dependency and Autonomy Costs. IEEE Transactions on Sustainable Computing 9(3), 315-328, May-June 2024; first published August 23, 2023. DOI: 10.1109/TSUSC.2023.3307790. Publisher abstract.