Innventure removes DarkNX AI data centre cooling project from bookings

September 09, 2026
Innventure told the SEC that the site named in Accelsius' DarkNX purchase order is no longer available and the project has been removed from bookings. In March the company had described the agreement as deployment across a 300MW AI data centre campus in Ontario.

Innventure has removed its subsidiary Accelsius' DarkNX cooling project from internal bookings after being told the deployment site no longer exists as planned, the company disclosed in its Form 10-Q filed on 13 August 2026.

The disclosure, under Item 5 Other Information, reads in full: "At this time, Accelsius is party to a purchase order with DarkNX whereby Accelsius is to provide cooling equipment and services concerning six, ten-megawatt deployments. Accelsius has recently been informed that the deployment site identified in the DarkNX purchase order is no longer available. Accelsius has removed the DarkNX project from its internal bookings pending identification of an alternative deployment location and satisfaction of other conditions."

Five months earlier, in a press release filed with the SEC on 4 March 2026, Innventure had described the same agreement differently, as one "with DarkNX to deploy NeuCool across a new 300MW AI data center campus in Ontario, Canada, expected to be the largest two-phase, direct-to-chip deployment to date." The August filing describes the purchase order as covering six ten-megawatt deployments, or 60MW. The filings do not reconcile the two figures, and a 60MW purchase order could represent a first phase of a larger campus.

On the same day as the 10-Q, Innventure suspended its 2026 targets for Accelsius. "Evolving dynamics in the AI infrastructure market, including constraints facing smaller early adopters around power availability, GPU access, and deployment timing, have impacted our near-term expectations," chief executive Bill Haskell said, adding that the company was "suspending our previously communicated expectations regarding Accelsius' 2026 revenue and cash flow targets."

Six days later, the board wrote to shareholders that senior management and directors who had received earnout shares on the basis of the Accelsius purchase order from DarkNX had agreed to forfeit them. The board said the shares "were issued properly based on contractual obligations established in 2023," but that forfeiture was appropriate "considering the subsequent removal of the DarkNX booking that satisfied the milestone."

The project had been publicly questioned before the disclosure. Morpheus Research, a short selling research firm, published a report on 28 May 2026 arguing that the DarkNX agreement was a fabrication and that DarkNX itself was a shell company, and said it found no evidence the campus existed. Morpheus disclosed in the report that it holds short positions in Innventure shares. Accelsius makes two-phase direct-to-chip liquid cooling systems, a technology aimed at the heat densities of AI server racks.

We hope you enjoyed this article.

Consider subscribing to one of our newsletters like Silicon Brief, AI Funding Brief or Daily AI Brief.

Also, consider following us on social media:

Subscribe to Silicon Brief

Weekly coverage of AI hardware developments including chips, GPUs, cloud platforms, and data center technology.

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more