XMax Inks 512-GPU AI Deal, Targets $100M Revenue

XMax Inc. (NASDAQ: XMAX) announced on November 2, 2026 that it entered a five-year artificial intelligence infrastructure services agreement with DSH HYPER PTE. LTD. of Singapore, giving it dedicated high-performance computing capacity deployed in Korea on infrastructure from kt cloud Co., Ltd. The company said the project targets 512 NVIDIA Blackwell Ultra B300 GPUs and expects more than US$100 million in cumulative revenue over five years. Service began October 29, 2026 with 480 GPUs live.

Context — Why the XMax AI Infrastructure Deal Matters Now

The agreement marks the first concrete buildout step for XMax's NeoCloud and AI Token Factory product lines, which the company has described as its AI growth strategy but had not previously tied to named infrastructure. XMax AI Inc., the company's wholly owned subsidiary, is the designated operating end user. DSH HYPER PTE. LTD. supplies the capacity; kt cloud, an affiliate of KT Corporation, hosts the physical infrastructure in Korea.

The company's own framing puts the cluster at the center of both product lines. NeoCloud is described as an AI-optimized cloud service for enterprise inference workloads. The AI Token Factory is an API-driven distributed inference platform that sells committed compliant capacity scoped by jurisdiction and compliance tier, aimed at finance, government, industrial and consumer goods customers.

XMax's corporate history gives the deal unusual weight. The company, formerly Nova LifeStyle, Inc., built its business on contemporary furniture design, sourcing and distribution, and only recently expanded into AI software and platform services through XMax AI and Elonx AI Holdings PTE. LTD. A US$100 million five-year revenue target from AI services is therefore a diversification claim rather than an incremental line item.

Closing conditions remain open. The five-year commitment depends on board approval and delivery of a security guarantee and standby letter of credit in favor of kt cloud by November 17, 2026. XMax said material terms will appear in a Form 8-K filed with the SEC on November 2, 2026.

The company did not disclose contract value, per-GPU pricing, or the identity of DSH's own suppliers. Those gaps matter because the revenue projection is tied to commercialization progress the company has not yet completed.

Data — What the 512-GPU Cluster Numbers Show

XMax's planned configuration is 64 compute nodes, each with eight NVIDIA Blackwell Ultra B300 GPUs, for a target total of 512 GPUs, plus 560TB of high-performance parallel storage. The initial live configuration is 480 GPUs, meaning roughly 94% of target capacity is already in service, with the full 512-GPU build expected in phases.

The revenue target of more than US$100 million is cumulative across five years, which implies an average of at least US$20 million per year once the platform is commercialized. The company said sources include AI inference billed on API usage, managed model services, and enterprise-grade AI application services.

Metric	Reported figure
Contract term	Five years
Target GPUs	512 NVIDIA Blackwell Ultra B300
Live GPUs at start	480
Compute nodes planned	64 (8 GPUs each)
Storage	560TB high-performance parallel
Revenue outlook	Over US$100M cumulative
SBLC deadline	November 17, 2026

The company did not provide a per-year revenue schedule, a margin estimate, or a customer count. Those omissions leave the US$100 million figure as a top-line target rather than a contracted backlog.

Analysis — Who Is Exposed to XMax's AI Pivot

XMax is effectively buying into the AI inference layer rather than the model-training layer. The B300 is a Blackwell-generation part, and the company said it intends to use the cluster primarily for high-throughput inference to optimize cost, latency and throughput. That positions XMax against established GPU cloud providers and inference API vendors, though the report names no competitors and no pricing benchmarks.

The counterparties carry their own exposure. kt cloud, an affiliate of KT Corporation, gains a named anchor tenant for Korean AI capacity. DSH HYPER PTE. LTD. sits between XMax and kt cloud as the contracting intermediary, a structure that concentrates delivery risk on a single Singapore entity the company did not describe further.

The clearest limitation is the conditionality. Board approval, the security guarantee, and the SBLC by November 17, 2026 are all prerequisites. If the SBLC is not delivered, the five-year commitment does not close on the stated terms. Currency movement between the Korean won and the US dollar is also flagged as a risk in the company's own forward-looking language.

Positioning is hard to read from public flow. XMAX is a small-cap Nasdaq listing whose prior identity was furniture, so the shareholder base is unlikely to overlap heavily with AI infrastructure specialists. The stock's reaction to the filing is not established by the report.

Outlook — Catalysts for XMax Through 2027

The nearest dated catalyst is November 17, 2026, the SBLC and security guarantee deadline in favor of kt cloud. Delivery of the SBLC, or disclosure that it was issued, is the first hard confirmation the agreement closes.

Second is the phased move from 480 to 512 GPUs, which the company said would follow in stages but did not schedule. Third is technical validation of the first AI Token Factory, which XMax said is its immediate focus, covering infrastructure integration, large-model inference optimization, model serving performance testing, and commercial platform validation.

Commercial launch of NeoCloud, MaaS and enterprise AI applications across the United States, Japan, Korea and Singapore is the revenue trigger. The company gave no launch date. Investors should treat the US$100 million figure as contingent on all three steps completing.

Frequently Asked Questions

What does the XMax AI deal mean for retail investors in XMAX stock?

It converts an AI strategy narrative into a dated infrastructure commitment. XMax now has a named supplier chain, a live 480-GPU configuration, and a five-year revenue target above US$100 million. It also introduces execution risk: board approval, an SBLC deadline of November 17, 2026, and platform validation all sit between the announcement and any revenue. The report gives no margin, pricing, or customer detail.

What is the difference between NeoCloud and the AI Token Factory?

NeoCloud is AI-optimized cloud infrastructure for enterprise inference workloads, sold as accelerated compute. The AI Token Factory is an API-driven distributed inference platform that is model-agnostic and deployment-agnostic, selling committed compliant inference capacity scoped by jurisdiction and compliance tier. The company plans to run Token Factory services on NeoCloud and add managed model endpoints and enterprise AI applications such as agents and retrieval-augmented generation.

Why does the standby letter of credit matter to the agreement?

The five-year service commitment is conditional on XMax's board approving the deal and delivering a required security guarantee and SBLC in favor of kt cloud by November 17, 2026. An SBLC is a bank-backed payment guarantee, so its issuance signals that XMax has secured credit support for the commitment. Without it, the agreement does not close on the stated terms, and the GPU capacity and revenue plan would not proceed as described.

Bottom Line

XMax's 512-GPU Korean AI buildout is real capacity, but the US$100 million revenue target depends on an SBLC due November 17.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.