SAIHEAT cedes 78% voting power in $60M Canopy Wave merger

SAIHEAT signed a definitive merger agreement with Canopy Wave on Aug. 10, valuing the AI inference startup at $60 million pre-money and giving its holders 78.44 percent voting control of the combined company.

"This combination will position the company where the AI market is going: inference at scale," said Jianwei Li, Chief Executive Officer of SAIHEAT. "Canopy Wave brings an inference platform and an exceptional engineering team."

The deal is structured through share issuance based on pre-money valuations of $60 million for Canopy Wave and $40 million for SAIHEAT. After closing and a concurrent private placement of about $4.5 million in Class A shares at $18.15 per share, former Canopy Wave stockholders will own approximately 54.19 percent of economic interests and 78.44 percent of voting power. The combined company, to be renamed Canopy Wave Holdings Inc., is expected to trade on Nasdaq under ticker CWAV.

The transaction effectively reverses SAIHEAT's public listing into Canopy Wave's platform, with the combined entity targeting the AI inference market for open-weight large language models. Closing is expected by end of 2026, subject to shareholder approval, Nasdaq listing approval, and completion of the private placement.

The ownership structure reflects significant dilution for existing SAIHEAT shareholders, whose economic interests will be reduced to approximately 45.81 percent. Canopy Wave founders Tao Zhang and James Liao, who will serve as CEO and CTO respectively, are expected to collectively hold a majority of the combined company's economic interests and voting power. The combined entity will relocate its headquarters to Santa Clara, California, and transition from foreign private issuer to domestic-issuer reporting requirements beginning next fiscal year. The boards of both companies unanimously approved the transaction. The deal structure means SAIHEAT's existing shareholders retain a minority economic stake while Canopy Wave's founders gain operational control, a dynamic that could influence how the combined company allocates capital between infrastructure and software investments.

AI Inference Spending Shifts From Training to Serving

The deal comes as enterprise AI spending rotates from one-time model training toward ongoing inference workloads — the process of generating model outputs, or "tokens," at production scale. Open-weight models have narrowed the capability gap with proprietary frontier systems from companies such as OpenAI and Anthropic, creating demand for cost-efficient inference infrastructure. Canopy Wave, founded in 2024, provides a full-stack platform with GPU cloud infrastructure, orchestration software, API endpoints, and security features including SOC 2 Type II certification and a zero-data-retention policy. The platform supports a broad catalog of leading open-weight models and serves developers and enterprises across AI coding, AI agent, and other production workloads.

SAIHEAT's modular data center business and energy-efficient computing capabilities are expected to complement Canopy Wave's GPU cloud operations, which currently rely on third-party infrastructure through leasing arrangements. The combined company will face competition from substantially larger providers in the AI infrastructure space, including Nvidia-backed cloud services and hyperscale data center operators. Tao Zhang, Canopy Wave's CEO, said the merger gives the startup a public-company platform and infrastructure depth to scale faster globally.

Financing and Execution Risks

SAIHEAT's effective F-3/A shelf registration dated July 10, 2026 permits offerings up to $300 million, providing financing flexibility for the combined entity. However, the transaction carries execution risks: closing depends on shareholder approval, Nasdaq's approval of the combined company's initial listing application, and satisfaction of conditions for the concurrent private placement. The combined company also faces exposure to export controls and trade restrictions given its dependence on third-party open-weight AI models, including models developed outside the United States. Canopy Wave's limited operating history since its formation in 2024 and customer concentration add further uncertainty.

The deal follows SAIHEAT's June 11 expansion into AI inference services for open-source models, which was met with a modest 0.7 percent share-price gain. The company's April strategy upgrade toward nuclear energy design software drew a 1.0 percent decline. SAIHEAT's market capitalization stood at $37.69 million with a float of 529,120 shares prior to the announcement. The combined company's success will depend on whether per-token pricing and GPU rental rates hold as competition intensifies across the AI infrastructure sector. The transition from foreign private issuer to domestic-issuer reporting will also bring increased disclosure requirements, potentially exposing the combined entity to closer scrutiny from U.S. regulators and investors.

This article is for informational purposes only and does not constitute investment advice.