The $2.4B AI Power Deal That Blew Up: What the Babcock & Wilcox Lawsuit Teaches About AI Hype

A $2.4 billion AI power contract sent Babcock & Wilcox shares up 198 percent. Then a short seller looked at who was actually on the other side of the deal. Now investors are suing, and the filing reads like a case study in how AI hype gets weaponized into a liquidity event for insiders.

The complaint (Cho v. Babcock & Wilcox Enterprises, Inc., Case No. 5:26-cv-00886, N.D. Ohio), filed April 14, alleges the NYSE-listed energy company and two senior executives sold the market on a massive AI data center power deal while leaving out the part where the company’s largest shareholder had its fingerprints on the counterparty. This is not a story about AI. It is a story about what happens when “AI” is the magic word that turns ordinary disclosure obligations into afterthoughts.

What the complaint actually alleges

The timeline in the complaint is tight and specific.

November 2025: Babcock & Wilcox announces a preliminary agreement to deliver one gigawatt of power to an AI data center operated by Applied Digital Corporation. CEO Kenneth M. Young calls the impact “profound” and says the deal adds over $3 billion to the pipeline. Project value: $1.5 billion plus.

The stock moves. Within days, the company launches an at-the-market offering and raises $67.5 million. By February 2026, shares are up more than 198 percent.

March 2026: The company announces full notice to proceed on a $2.4 billion design-build contract with Base Electron, described as a newly formed independent power producer backed by Applied Digital. Reported backlog jumps 470 percent. The stock adds another 45 percent in a single session.

March 12, 2026: Wolfpack Research publishes a short report questioning the whole structure. Shares drop 11.59 percent that day.

The complaint’s core allegation is that Base Electron was not the arm’s-length counterparty investors thought it was. According to the filing, BRC Group Holdings (formerly B. Riley Financial), which was Babcock & Wilcox’s largest shareholder, had direct ties to Base Electron. BRC’s co-CEO and chairman, Bryant R. Riley, was allegedly a director of Base Electron. Base Electron’s registered address allegedly matched BRC’s headquarters, not Applied Digital’s. And Base Electron was allegedly not even incorporated until seven weeks after the preliminary deal was first announced.

While the stock was riding the AI narrative, BRC allegedly sold its entire directly held stake in Babcock & Wilcox for about $10.4 million at prices 140 percent above the pre-announcement level.

The Wolfpack report also flagged that Applied Digital’s existing data centers already had power through conventional grid agreements, raising the question of whether this contract would ever produce real revenue. And the complaint notes that Applied Digital could walk away from its guarantee of Base Electron’s obligations for as little as $50 million.

These are allegations, not findings. Nothing has been proven. But the shape of the story is familiar enough that general counsel and compliance teams should be paying attention.

Why this matters beyond one stock

Every GC at a company with an AI narrative is watching this case, whether they know it or not. Three reasons.

First, related-party disclosure is suddenly back in the spotlight. When AI is the magic word that drives a 198 percent move, the question of who is on the other side of your AI deal becomes material in a way it might not have been for a routine customer contract. If your largest shareholder is sitting on both sides, that needs to be in the release, not buried in a 10-K two quarters later.

Second, “pipeline” and “backlog” numbers tied to AI projects are going to get sharper scrutiny. A 470 percent backlog jump on a single contract with a newly formed counterparty is the kind of number that short sellers build entire reports around. If your company is putting out AI-linked numbers that depend on counterparties that cannot easily be verified, expect the complaint to cite those numbers verbatim.

Third, the ATM offering matters. Raising $67.5 million during the run-up, before the full structure is disclosed, is exactly the kind of fact pattern plaintiffs’ firms build securities fraud complaints around. The timing of capital raises relative to disclosures is going to be one of the first things discovery goes after.

The short seller problem is not going away

Wolfpack Research is not a regulator. It is a private firm with a financial position. But the playbook here (publish a detailed report, trigger a single-day drop, watch the class action get filed within weeks) is now a reliable engine for securities litigation in AI-adjacent names. The short seller does the forensic work. The plaintiffs’ firm picks up the report, files the complaint, and lets discovery do the rest.

For boards, that means any company with a big AI narrative needs to assume that somebody is building a file on every counterparty, every press release, every insider trade. The time to fix a messy related-party disclosure is before the short report drops, not after.

What the SEC is almost certainly watching

The SEC has been public for more than a year about “AI washing” as an enforcement priority. Cases involving specific, material misstatements tied to AI deals (not generic “we use AI” language) are the ones most likely to draw a parallel investigation. If even half of the Cho complaint’s allegations hold up, the SEC will have a hard time leaving this one to the private bar.

The near-term question is whether the Ohio federal court survives a motion to dismiss. The long-term question is whether the SEC files a parallel action with specific findings about related-party concealment. Either outcome sets precedent for how AI-linked disclosures get scrutinized going forward.

What to do now

If you are a general counsel or compliance officer at a public company with an AI story, use this case as a forcing function this week.

1. Map every material AI-related contract to its counterparty’s ownership and control. If any counterparty has overlapping ownership with a major shareholder, officer, or director, document it and make sure disclosure counsel has signed off.

2. Audit your last four quarters of AI-related press releases for language that implied arm’s-length counterparties where the structure was actually more complicated. If anything looks soft, fix the 10-Q narrative before the next filing.

3. Review your insider trading windows against the timing of AI deal announcements and capital raises. If a 10b5-1 plan executed inside a hot window, have counsel refresh the file.

4. Lock down how “pipeline” and “backlog” numbers get calculated and who signs off. The days of a CEO quote adding billions to the pipeline on a preliminary agreement are ending.

5. If you are on the investor side, assume that every AI-linked contract announcement with a round-number valuation and a new counterparty deserves a read of the counterparty’s formation documents. Base Electron was not incorporated until seven weeks after the preliminary deal. That was in the public record. Somebody just had to go look.

The Babcock & Wilcox case may settle, may get dismissed, may go to trial. The lesson is already banked. AI does not immunize a deal from the disclosure rules that applied before AI existed. If anything, the magic-word effect on stock prices makes the disclosure obligations more important, not less.