iPower AI signs five-year LOI targeting $21.9m in annualised services
iPower Inc. (Nasdaq: IPW) has announced that its wholly owned subsidiary, iPower AI LLC, has received a customer-signed, non-binding letter of intent covering a five-year AI equipment services and management relationship. The deal represents the most concrete commercial step taken by the California-based company since it began repositioning itself as an AI infrastructure and services provider.
Under the proposed terms, the unnamed customer has identified an initial fleet of 48 AI computing units, with plans to scale to approximately 396 units during 2027. At the proposed fee of $4,608 per unit per month, the full build-out would generate up to $1.82 million in monthly service fees, equating to up to $21.90 million on an annualised basis. The initial 48-unit phase would represent roughly $2.65 million in annualised fees, or $13.27 million if all units remain billable for the full 60-month period.
iPower AI LLC would provide or coordinate equipment operations, monitoring, maintenance, technical support and data-centre management on behalf of the customer. The model is explicitly designed to be asset-light: the company says the services relationship is not limited to infrastructure that iPower itself owns, opening the path to a broader managed-services play.
Lawrence Tan, chief executive of iPower, said the agreement "puts a concrete operating model" behind the company's AI pivot. "We are working to make AI infrastructure and services a central driver of what iPower becomes," he added.
Caveats are substantial
The release carries significant qualifications that readers should weigh carefully. The LOI is explicitly non-binding, and commercial terms remain subject to a definitive services agreement. The revenue figures cited are illustrative projections that assume units remain billable at the proposed rate for the full contract period. iPower states directly that the figures do not constitute committed revenue, backlog or 2027 guidance.
The customer is not named, no definitive agreement has been signed, and there is no assurance one will be. The forward-looking statements section in the release is unusually detailed in its risk disclosures, reflecting the speculative nature of the announcement.
Market context and competitive position
The managed AI infrastructure services market is crowded and increasingly stratified. At the top end, hyperscalers including AWS, Microsoft Azure and Google Cloud dominate enterprise compute; below them, a tier of specialist GPU cloud providers such as CoreWeave, Lambda Labs and Voltage Park competes on price, flexibility and customer proximity. Smaller operators are attempting to carve out positions in co-management and white-label infrastructure services, targeting customers who own or lease hardware but lack in-house operational capability.
iPower's proposed fee structure, at roughly $4,600 per unit per month, sits in a range consistent with managed services mark-ups on GPU or AI accelerator hardware, though the release does not specify the underlying hardware type or supplier. Without named customers, disclosed hardware specifications or a signed contract, the LOI's commercial value cannot be independently assessed at this stage.
The company's pivot from e-commerce and hydroponic equipment into AI infrastructure, signalled in its current SEC filings, is still at an early stage. Investors will be looking for a signed definitive agreement and disclosure of the hardware vendor and data-centre partner as the minimum evidence that the operating model is progressing beyond the letter-of-intent phase.