Maase signs RMB14.76m AI computing deal with Guangzhou Chuangyan
Maase Inc. (NASDAQ: MAAS) has announced that its subsidiary Huarong Future (Sichuan) Technology Co., Ltd. has signed a computing services agreement with Guangzhou Chuangyan Information Technology Co., Ltd., a Guangzhou-based technology company. The contract runs for 12 months at a fixed monthly fee of RMB1.23 million, producing a total contract value of RMB14.76 million (approximately $2 million at current exchange rates).
Under the agreement, Huarong Future will deliver 90P-class FP16 half-precision floating-point compute capacity along with technical consulting, support and hardware-layer troubleshooting. The contract specifies a minimum technical threshold: the aggregate theoretical peak FP16 performance of the supplied compute cluster must exceed 90 petaflops under non-sparse computing conditions. Maase did not disclose the underlying GPU hardware, the data-centre location, or whether capacity is proprietary or sourced from a third-party cloud provider.
The deal
The agreement moves Maase toward what its chief technology officer, Dr. Zhifeng Li, described as a shift from project-based delivery to a recurring computing services model. "As AI model training, AI inference and enterprise AI applications continue to expand, demand is growing for computing resources that are stable, scalable and supported by ongoing technical services," Li said. "MAAS will continue to strengthen its capabilities in computing resource integration, technical delivery and operational services."
Guangzhou Chuangyan is not named as an existing customer in prior Maase filings reviewed for this brief, and no detail was provided on Chuangyan's end use case, whether AI model training, inference serving, or another application. The deal is structured as a managed capacity contract rather than a software licence or platform subscription, which means revenue recognition is tied to monthly service delivery rather than an upfront booking.
Market context
The contract arrives as demand for contracted GPU compute capacity in China continues to intensify, driven by domestic AI model development and a wave of enterprise AI application deployments. Chinese firms operating in this segment face a complex supply environment: US Bureau of Industry and Security export controls imposed in late 2023 and tightened in 2024 restrict access to the most advanced Nvidia accelerators, pushing domestic operators toward locally sourced alternatives such as chips from Huawei Ascend and Cambricon, or older-generation Nvidia inventory where it can be obtained.
Maase, which positions itself as an AI-centric full-scene digital systems provider, competes in a fragmented market that includes larger state-backed cloud operators, independent GPU leasing platforms and hardware resellers offering managed services. The company's ability to sign recurring services contracts rather than one-off hardware deals would, if sustained, improve revenue visibility and margins relative to project-based delivery. Investors will be watching for contract renewal rates and any expansion of the Guangzhou Chuangyan agreement as indicators of commercial traction.
Regulatory read-across
As a NASDAQ-listed, China-headquartered technology company, Maase files with the US Securities and Exchange Commission and is subject to the Holding Foreign Companies Accountable Act audit-access requirements. The forward-looking statements section of the release explicitly flags the risk that the agreement may not be fully performed or renewed, and that contracted revenue may not be fully recognised. Enterprise buyers and investors should treat the RMB14.76 million figure as a contractual ceiling rather than a guaranteed revenue booking.
The 12-month structure and monthly-fee model suggest Maase is building a service layer above raw infrastructure, a positioning that could support higher gross margins over time but requires consistent uptime and technical support delivery to underpin renewal.