What public companies just told the SEC about AI, read straight from the filings. Every quote links the original document.
XMax Buys GaN Chip Maker Targeting AI Power Infrastructure
XMax Inc. (NASDAQ: XMAX), which describes itself as a furniture distributor and AI software provider, announced a definitive share purchase agreement to acquire 100% of Hexa Creation Inc., a U.S. company developing 1200V vertical gallium nitride (GaN) power semiconductors. The deal is framed explicitly around AI data center power infrastructure. GaN at 1200V targets the high-power density requirements of GPU clusters, where conventional silicon power devices face efficiency limits at high voltages.
Transaction gives XMax footing in high-voltage GaN power technology aimed at AI data center power infrastructure
The acquisition is disclosed in an 8-K exhibit, meaning XMax considered it material enough for immediate reporting. A furniture and AI software company acquiring a power semiconductor developer is an unusual combination, and the filing makes no attempt to explain the strategic fit beyond the AI data center angle.
Quanome Signs Enterprise Contract for Dedicated AI Inference Capacity
Quanome Technologies (Nasdaq: QNME) reported in an 8-K that its subsidiary XDT Infrastructure I, LLC executed a Master Services Agreement and initial Order with an unnamed enterprise customer for dedicated AI inference capacity. The service is defined plainly in the filing: the customer runs trained models against new data through XDT's infrastructure without building its own.
Rather than purchasing and operating its own specialised AI infrastructure, the customer will access the service remot…
The structure separates inference from model training, with Quanome supplying the inference layer as a managed service. The filing does not disclose the dollar value of the initial Order.
AI Named a Cyberthreat Amplifier in Sports Spinoff Filing
MSGS Spinco, Inc. (MSGR), spinning off from Madison Square Garden Sports, used its 8-K exhibit to describe AI as accelerating cyberattacks in two specific ways: faster, more sophisticated threats, and easier discovery of exploitable vulnerabilities. That pairing is more precise than the generic AI-creates-new-risks language typical of risk sections.
The rapid development and increased availability of advanced artificial intelligence technologies may further heighten these risks by increasing the speed, scale and sophistication of cyber threats and making security vulnerabilities easier to identify and exploit.
Clinical-stage biotech Bambusa Therapeutics, filing its S-1 in the same window, flagged that generative AI use by employees creates new exposure under privacy laws and that further AI regulation is expected. Two companies in unrelated sectors landing on the same dual-risk framing in the same window suggests this language has reached standardization across filing types.
S-1 Queue: Quantum-AI Integration, an AI Cancer Lab, and B2B Data Creation
Quantum computing startup SeeQC (SEQC) filed its S-1 describing its architecture as enabling tight integration between quantum processing units and classical compute including GPUs. The company names as its target workloads both AI applications running on quantum hardware and quantum applications augmented by AI processing, framing the two as mutually reinforcing rather than competing.
DataMeds AI (MEDS) disclosed in its S-1 amendment that it closed the Helomics acquisition on September 11, 2026, funded partly by a $1,363,672 convertible note bearing 7% annual interest due in 2029. Helomics is now a wholly owned subsidiary.
Helomics operates an artificial intelligence cancer diagnostics laboratory business.
Mobile content platform Zedge (ZDGE) filed an employment agreement that formally defines its business to include multimodal data creation services for AI developers and enterprises. Embedding that definition in a binding employee IP agreement indicates Zedge treats the AI data-services line as a core asset requiring contractual protection.