EL7.AI
StocksMedium•26 October 2026•
3 min read

$94.4 Million Deal Reserves 96 AI Servers in Dallas

Key Facts

1Agreement reserves Nvidia B300 capacity for 36 months at the Dallas data center.
2The reservation covers 96 enterprise AI servers.
3The contract has a total value of about $94.4 million over 36 months.
4The deal represents the joint venture’s first contracted customer commitment.
5The company expects to deploy GPU units in Dallas in late December 2026 or early January 2027.

Forum Edge AI LLC, the joint venture between Forum Markets and Edge Node Inc., has agreed with Fireworks to reserve computing capacity powered by Nvidia B300 units for 36 months at the venture’s Dallas data center, with a total contract value of about $94.4 million. The reservation covers 96 dedicated AI servers, according to a filing with the U.S. Securities and Exchange Commission. The value includes GPU capacity fees and CPU management contract fees over the term of the agreement. Forum said the deal represents the venture’s “first contracted customer commitment” and its first definitive contract. GPU units are expected to be deployed in Dallas in late December 2026 or early January 2027.

The agreement reserves capacity for a set period, while the expected operational deployment begins weeks after signing, according to the timeline announced by the company. The filing did not specify a single deployment date, instead setting a window from late December to early January. It also states that Fireworks paid an upfront deposit to be credited against fees due during the contract term. The filing did not disclose the deposit amount, so the portion of the total contract value paid upfront cannot be determined. The deal links existing data center capacity in Dallas with contracted demand for computing services, with the pace of operations to become clearer once the units are deployed.

Forum’s chief executive described the agreement as “an important validation of our AI infrastructure strategy,” citing the company’s focus on rapidly deploying revenue-generating GPU units through a modular approach. The contract is significant as the venture’s first contracted customer commitment; the venture was established to develop AI computing capacity across existing sites. When Forum announced the joint venture, it said its initial portfolio included a data center in Dallas, a data center campus in High Point, North Carolina, and an initial site for a mobile communications tower. This deal concerns Dallas capacity, while the filing does not link the contract to the North Carolina site or the communications tower. The agreement thus provides the venture’s first announced definitive contract under its broader plan to make use of existing infrastructure sites.

Fireworks provides inference, training and model-routing services, and says it helps enterprises run and scale generative AI applications in production environments. Fireworks said its services enable companies such as Uber and Shopify to train and run custom models. Forum valued Fireworks at about $17.5 billion in its latest funding round. The filing named Nvidia, Atreides Management, Index Ventures and TCV among the company’s backers. It also said Hosted AI Inc. facilitated the relationship between Forum Edge AI and Fireworks. These details place the customer in the context of a service provider that supports enterprise model deployment, while the announced contract centers on reserving server and GPU capacity in Dallas.

Forum had announced 2027 consolidated revenue guidance of between $125 million and $175 million, and said it expected to turn cash-flow positive early that year. These are company guidance figures, not realized revenue or proceeds disclosed from the Fireworks contract alone. Forum linked that guidance to assumptions about the timing of site execution, including the Dallas electrical transformer upgrade, the closing of the North Carolina site and the pace of installation of the first tower. It also warned that power, permits, equipment or contracts could be delayed. The expected GPU deployment date in late December or early January falls within the timeline the company will monitor as it brings the contracted capacity online, alongside the other site execution milestones cited in its guidance.