Super Micro forecast $72 billion. The number may be irrelevant.

Super Micro has told the SEC that its internal controls do not work, in every quarterly report it has filed for more than a year. A forensic accountant who has spent more than thirty years reading filings like these was blunt about what that means:

The length of time it takes to remediate control weaknesses is a good proxy for how ineffective the system was when the problems were discovered. This case is unusual in that the company has had two major periods of internal control weaknesses: back in the 2015-2020 period then again more recently. If I had to bet, controls were probably ineffective in between as well. I can't say I've ever seen a company that has had internal control weaknesses for the better part of a decade. That's a huge red flag.

Donn Vickrey co-founded Gradient Analytics and Pacific Square Research, both forensic research firms, and an earlier venture of his, Equity Methods, was bought by Bank of America. He taught accounting at the University of San Diego and published on earnings quality in the Financial Analysts Journal. He has not looked closely at Super Micro in some years, he said, because his firm avoids battleground stocks, so what follows is his read of the pattern rather than a company-specific call.

On 11 August, Super Micro told investors to expect between $65 billion and $72 billion of revenue in fiscal 2027. Analysts had been expecting about $54.4 billion. Revenue for the quarter that had just ended actually came in below expectations, but the shares rose about 13 percent anyway, carried by a gross margin far above what the company had guided.

The forecast is produced by a company whose chief executive and chief financial officer have concluded, quarter after quarter, that its disclosure controls do not work.

On the forecast itself:

How much weight should be placed on revenue estimates? I'd say very little. Wall Street expectations will be what drives the market's response. And when you have significant internal control weaknesses the risk of missing is substantially greater.

What the company says about itself

The language is in the most recent quarterly report, and it is not ambiguous. The chief executive and the chief financial officer concluded that disclosure controls and procedures "were not effective at the reasonable assurance level as of March 31, 2026" because of material weaknesses in internal control over financial reporting.

Those weaknesses were not new that quarter. They were first identified in the annual report for the year ended 30 June 2024, which was filed late, on 25 February 2025. They were still listed as unremediated more than a year later. There are four of them.

Where he said to look, and what the filing says about it

Asked what he would examine instead of the forecast, Vickrey pointed to one thing. He said he would "attempt to identify and eliminate sales to related parties, especially those that could be responsible for sales of prohibited chips to China." That, he said, "would not be easy, but it's probably the crucial missing piece for estimating 26 and 27 revenues."

Read the fourth material weakness again. The control the company says it has not designed, implemented or documented is the one covering "the timely identification and disclosure of new related party transactions." The single number an outside analyst would most want in order to test the forecast is the number the company has told the SEC it cannot reliably produce.

The China element is not hypothetical either. On 19 March 2026 the US Attorney for the Southern District of New York returned an indictment concerning three individuals employed by or associated with Super Micro at the time, in connection with an alleged conspiracy to commit export control violations. On 7 April the company announced an independent investigation into those allegations, led by its lead independent director and the chair of its audit committee. The filing says that investigation is ongoing and that no conclusion has been reached. The three individuals are not the company, and nothing has been proven.

Twenty-one months of "unable to estimate"

Super Micro received an SEC subpoena on 19 November 2024 and a second one on 28 April 2026. In the quarterly report the company says of the investigation, as it does of the securities class action and the derivative suits, that the matter is "too preliminary to form a judgment" and that it is "unable to estimate the possible loss or range of loss, if any."

Asked whether that language can hold indefinitely, Vickrey said it usually does:

The "unable to estimate" language is common and I can't say I've ever seen a company change that statement until the amount of the restatement is determined. The fact that there is an SEC investigation and a DOJ investigation make this one particularly problematic. It could take quite a while and it's really not easy to know how bad it will get for them in terms of losses/restatements.

His view is that the eventual dollar figure is not really the point. The losses themselves, he said, will likely be ignored by the market and are useful mainly as a proxy for how bad the underlying situation turns out to be for future sales.

"The forecast is irrelevant"

Asked directly how he would assess the $65 billion to $72 billion range if he could not rely on the controls behind it, Vickrey did not treat it as a modelling problem:

Honestly, this is a company I'd avoid like the plague. The forecast is irrelevant IMO. The same CEO has been in charge during both periods of alleged misstatements and other alleged wrongdoing. I just don't invest in companies with that level of regulatory baggage. There are plenty of companies that don't have that baggage.

That is a personal investing position rather than a finding, and it should be read as one. It is also the view of somebody who has spent three decades reading these filings for a living, and it is a good deal blunter than anything on the sell side.

The year the controls produced

It is worth looking at what those systems reported for fiscal 2026, because the numbers are strange in their own right. Revenue rose 78 percent to $39.1 billion and net income more than doubled to $2.23 billion. Operating cash flow was negative $6.81 billion, against positive $1.66 billion the year before.

The gap sits in working capital. Inventories rose 176 percent to $12.9 billion and receivables rose 178 percent to $6.1 billion, both growing more than twice as fast as sales. The company covered the shortfall with $9.48 billion of net financing inflows, raised from term loans, a mandatory convertible preferred issue and common stock. Cash on the balance sheet went up, which is why it does not look like distress, but none of it came from operations.

One more thing about that fourth quarter. Gross margin was 9.9 percent in the third quarter and 17.5 percent in the fourth, close to doubling in the final three months of a fiscal year. Management attributes it to a richer enterprise customer mix. That is the company's explanation, and it is also the quarter that has not yet been through an audit.

What to watch

The fiscal 2026 annual report is due within weeks and is the first document that matters here. It will say whether any of the four material weaknesses have finally been remediated, and it carries the related party transactions note, which is where Vickrey's question can actually be tested against a number rather than argued about.

The second thing is the independent investigation into the export control allegations, which the board said would report to its four other independent directors. The third is simply whether operating cash flow turns. The fourth quarter produced $747 million of it, so the burn is concentrated in the first nine months, and one good quarter is not yet a trend.

Until the annual report lands, the position is the one Vickrey described. A company with an unusual history of broken controls has issued a forecast eleven to eighteen billion dollars above what analysts expected, and the control most needed to check it is, on the company's own account, not working.

Cover photograph: server racks, by NOIRLab/NSF/AURA/T. Slovinský (opens in a new tab), via Wikimedia Commons, licensed CC BY 4.0 (opens in a new tab). It shows a generic server room rather than Super Micro hardware. Cropped and resized from the original.