A $50 Million Deal With a Company That Has $2 Million

A four-month-old Chinese company with $2 million in startup capital agreed to a $50 million deal to buy artificial intelligence chips. That was the headline claim. It was also the thing that started seven law firms racing to represent the same class of angry investors.

The buyer was NeoTensr. The seller was Blaize Holdings, a programmable AI chip company that went public through a SPAC merger in January 2025. Blaize announced the deal on April 16, 2026. The stock jumped roughly 25 percent. Two weeks later, on April 28, a short seller called Pelican Way Research published a report arguing the deal was basically a cover for a capital raise. The stock fell to $1.90 and has not looked back.

The simplest model is this: if your company burns roughly $74 million a year and you have about $46 million in the bank, you need fresh capital in roughly six months. One way to get it is to raise equity at whatever the market will pay. Another way is to announce a big deal first, push the stock up, and then raise the equity at a better price. The Pelican Way report's title made that connection explicit: "Blaize AI: Running Up The Share Price Based on a Seemingly Bogus Deal, Conveniently Timed for Massive Dilution."

That is the weird fact you have to start from, because it frames everything else.

Here is the sequence. Blaize announced the NeoTensr partnership on April 16 — a contract for co-branded edge AI infrastructure worth up to $50 million. The stock soared. On May 5, Blaize announced a public offering of common stock. It priced the next day at $1.85 per share, selling 18.9 million shares and raising about $35 million. That was 13 percent of the roughly 142 million weighted-average shares outstanding, according to Blaize's own Q1 filing. The dilution was real. The timing was what Pelican Way called convenient.

But the deal itself is where the mechanics get stranger. According to the short report, NeoTensr's website was registered on December 18, 2025, and the company incorporated four days later on December 22. Chinese entity filings showed approximately $2 million in startup capital. Blaize, for its part, said in April that it had already recognized $20 million in revenue from NeoTensr during the fourth quarter of 2025 — meaning NeoTensr had reportedly ordered $20 million of product before it was even a legal entity, or in the very weeks around incorporation.

The complaint language from the class action filings is blunt: Blaize "created the appearance of growth by announcing transactions with entities not capable of meaningful business activities". The companies and executives that signed those deal announcements allegedly knew the counterparty was "wholly unequipped to conduct meaningful business."

This isn't the first time Blaize has been accused of something like this. Pelican Way references a pattern. In 2025, Blaize announced a roughly $120 million deal with another Chinese entity called Starshine Computing. As of March 24, 2026, according to Blaize's own 10-K filing, Starshine had paid $1.6 million of what appears to have been an $8.8 million accounts receivable balance. The remainder was still outstanding. That is less than 20 percent collection on a deal the market was apparently supposed to treat as a sign of traction.

The basic point is that you don't need to prove fraud to understand the incentives. A cash-burning company with limited revenue — Blaize reported $38.6 million total in 2025, with nearly all of it coming from two customers apparently located in China — needs to show the market that something is going right. Announcing a large partnership does that. If the partnership also happens to lift the stock price right before an equity offering, it does even more. Whether the deal was "bogus" in a legal sense, or just optimistic in a way that didn't survive contact with Chinese filing records, is a question for the complaint. The structural logic is visible without resolving it.

Then there's the rights plan. On April 22, 2026 — six days after announcing the NeoTensr deal and two weeks before Pelican Way's report — Blaize adopted what it called a "limited duration stockholder rights plan". In plain language, that is a poison pill. It is designed to prevent any single entity or coordinated group from acquiring 10 percent or more of the company's shares without the board's blessing. The company said it adopted the plan in response to "a specific threat of stockholders seeking to form a group capable of exerting negative or actual control without appropriately compensating all stockholders." That is the respectable phrasing. The economic reality is that a rights plan gives the board more room to negotiate the terms of the next capital raise, because it limits the leverage that an activist or a large investor can bring to the table.

The sequence — big deal announcement, poison pill, equity offering, short report, stock collapse — reads like a company under liquidity pressure trying to manage the terms of its next funding round. Whether you think that's fraud or just aggressive capital management depends on whether NeoTensr was ever capable of paying $50 million. I don't know the answer to that question. The complaints don't include an admission from Blaize, and no court has ruled on anything yet. But the timeline does the work of explaining why seven law firms are filing on the same class period — July 18, 2025 through April 28, 2026 — and why they're all asking the same investors to contact them by October 5 to apply for lead plaintiff status.

The Rosen Law Firm press release that triggered this story is itself a routine artifact of the securities class action ecosystem. Rosen joins Schall, Brown & Schwartz LLP, the Law Offices of Howard G. Smith, Pomerantz LLP, Glancy Prongay Wolke & Rotter LLP, Faruqi & Faruqi, and Bronstein, Gewirtz & Grossman in circling the same complaint, the same class period, and the same pool of potential plaintiffs. That's not a sign of exceptional fraud. It's the standard competitive dynamics of fee-driven litigation. Multiple firms file on the same case; the court picks a lead plaintiff and, usually, that plaintiff's counsel. The rest of the firms get nothing. The race to the filing window is the business model.

So if you're a BZAI investor who bought between mid-July 2025 and late April 2026 and lost money, you may have a claim. The stock is at $1.10 now, down roughly 44 percent year-to-date and 66 percent on a rolling annual basis. At that level, the math of a class action recovery is interesting even for the lawyers. But the class action is downstream of the actual story.

The actual story is about what happens when a SPAC-era AI hardware company, burning through cash faster than it can generate revenue, needs to keep the market's attention long enough to raise more capital. The NeoTensr deal may have been real. It may have been paper. It may have been somewhere in between — a legitimate partnership with a counterparty that simply didn't have the balance sheet the press release implied. But the timing of the announcement, the rights plan adoption, the equity offering two weeks later, and the Pelican Way report that connected all three — that sequence is the plumbing. The lawsuits are just what happens after the plumbing gets inspected.

The classification boundary that matters here isn't between fraud and optimism. It's between a company whose business model can actually work at the unit economics it describes, and one that needs the market to believe in the pipeline before the pipeline can exist. Blaize told investors in its SPAC prospectus that it had raised over $330 million from strategic investors including Denso, Mercedes-Benz, Magna, Samsung, and Temasek, with a qualified pipeline valued at more than $400 million. That's a respectable pedigree. But pedigree doesn't solve the funding model question. A company that burns $74 million a year needs to either slow the burn or raise capital at a price that doesn't destroy existing shareholders. Both paths require credibility. Credibility gets harder to maintain when your largest announced deals come from four-month-old counterparties with $2 million in capital.

The lead plaintiff deadline is October 5. The court hasn't certified anything yet. Blaize hasn't admitted anything. But the sequence of events — and the timing between the deal announcement, the rights plan, the equity offering, and the short report — is visible in public filings. That's what the seven competing law firms are all pointing at. The question for investors isn't whether they should file a motion. It's whether a company that can announce $50 million deals with entities that appear to have no capacity to pay them is running a revenue machine or a capital-raising machine. The stock price has already answered that question. The lawsuits are just trying to monetize the answer.