SMCI Stock Rockets 20% As Gross Margin Guidance Almost Doubles

On the evening of July 21, 2026, Supermicro told investors something it had not been able to say in over a year: profits were about to look a lot healthier. The AI server maker had already closed the regular trading session up 7%, at $25.50. Then, after the bell, came the real fireworks. 

In a routine-sounding "business update," the company said its gross margin guidance was jumping from a wafer-thin 8.2% to 8.4%, all the way up to 15% to 17%, backed by more than $60 billion of new orders booked in a single quarter, a company record. SMCI shares spiked as much as 20% in after-hours trading, touching the $30 mark. For a stock that has spent the better part of two years synonymous with an accounting scandal, an auditor's resignation, and a brush with getting delisted from the Nasdaq, that is not a small move. It is the kind of night that either marks a genuine turning point, or one more swing in a stock that has made a habit of them.

Let's break down exactly what Supermicro announced, why a margin number jumping from single digits to mid-teens matters this much for a hardware business, and whether a stock still down roughly 75% from its all-time high deserves a second look, as a value bet, a growth bet, or neither.

Why Is SMCI Stock Rising?

This came straight from Supermicro's own investor relations page, filed as a business update for the fourth quarter of fiscal year 2026 (the company's fiscal year runs through the end of June, so this covers April through June). The exact words are worth reading closely: "Backlog rose to record levels at the end of fiscal 2026 with total new orders in excess of $60 billion received during the fourth quarter of fiscal 2026." On margins, the company said gross margins were estimated at 15% to 17%, "significantly higher than our guidance of 8.2% to 8.4%, primarily due to a favorable customer and product mix."

Revenue, interestingly, was not the star of this update. The company said sales would land near the low end of its already-issued guidance, so the entire surprise here was about future profitability and order flow, not this quarter's top line. Every version of the story agrees this was one of Supermicro's sharpest single-session moves in months, and it pulled Dell and HPE up 5% to 6% after hours too, since both compete in the same AI server business.

One thing worth flagging before we go further: Supermicro was explicit that these are preliminary, unaudited numbers. Its auditor, BDO, has not reviewed any of it yet. The company also repeated that its board is still conducting an independent review connected to alleged export control issues, something we will come back to. Full audited results arrive with the August 11, 2026 earnings call.

Why Supermicro’s $60 Billion in New Orders Matters

Backlog is a simple idea once you strip the jargon away. It is the value of confirmed orders not yet delivered, a bit like a wedding caterer already booked solid through the next two wedding seasons even though none of those events have happened yet, and none of that money has been collected. The size of that pipeline still tells you a lot about how much work is coming.

For Supermicro, $60 billion of new orders landed in one quarter alone. That is more than one and a half times the company's entire expected revenue for all of fiscal 2026 (guided between $38.9 billion and $40.4 billion), and nearly three times what it actually booked as revenue in the whole of fiscal 2025 ($21.97 billion).

Backlog is a promise, though, not a payment, and Supermicro's own filing says so directly: some of these orders "may not constitute firm commitments and may be subject to cancellation or delays." A useful way to size this kind of number is what we would call a Backlog Coverage Ratio, simply new orders divided by trailing annual revenue. A ratio near 1x suggests roughly a year of extra visibility already locked in. Supermicro's Q4 figure works out to a ratio of about 1.5x against its own FY2026 guidance, an unusually high number for a hardware company, and a big part of why the market reacted so strongly even though this quarter's actual revenue was nothing special.

Cracking The Code On "Favorable Customer And Product Mix"

That phrase is doing an enormous amount of work in this announcement, and it deserves more than a skim.

Supermicro does not sell one product at one margin. It sells everything from bare-bones rack servers to fully assembled, liquid-cooled AI clusters built around Nvidia's latest chips, and margins on these vary hugely depending on who is buying and how the deal was priced. Selling a commoditized rack to a giant cloud company squeezing every vendor for the lowest price looks nothing like, margin-wise, selling a complex, hard-to-replicate liquid-cooled system to a customer that values Supermicro's engineering as much as its discount.

Think of it the way a restaurant thinks about its menu. A thali counter serving high volumes at a fixed, competitive price survives on turnover, with wafer-thin margins per plate. A kitchen doing customized, made-to-order dishes for a smaller set of diners can charge more per plate, because it is selling expertise and customization, not just calories. When Supermicro says its mix turned "favorable," it likely means more of this quarter's business leaned toward the customized, higher-value end of its menu instead of the high-volume, low-margin end.

The company has not broken down exactly which customers or products drove this, and that gap is worth remembering. A margin jump from a genuine shift in the customer base is a very different story from one driven by a single, unusually profitable one-off deal. Only the fuller results and management's own commentary on August 11 will tell us which one this actually is.

Why Is SMCI Stock Still Down 75% From Its Record High?

Context matters, and Supermicro's context is unusually dramatic even by AI-stock standards.

The stock's all-time high, adjusted for a 10-for-1 split the company carried out in 2024, was an intraday $122.90, hit on March 8, 2024, right around when S&P Dow Jones Indices announced Supermicro would join the S&P 500. CNBC noted at the time that the stock had already run up more than 200% since the start of that year and roughly twentyfold over the prior two years. Measured against the July 21 regular-session close of $25.50, the stock is down about 79%. Measured against the after-hours price near $30, the decline is closer to 75%.

A drop that size rarely comes from one bad quarter. It came from a run of governance crises stacked one on top of another.

Two entries deserve extra attention. The EY resignation was the scariest single data point for investors, since auditors rarely walk away from a client mid-relationship, and when they do, it is almost always about trust, not paperwork. EY's own language was blunt: it said it could no longer rely on management's and the audit committee's representations, and did not want its name attached to the numbers Supermicro was publishing.

The special committee that investigated afterward is the more reassuring entry. It found no evidence of fraud or intentional misconduct, disagreed with EY's stated reasons for quitting, and did not require Supermicro to restate its prior financial statements. It did, however, flag real weaknesses in internal controls, weaknesses serious enough that new auditor BDO has issued an adverse opinion on those controls in each of the two years since. Passing the fraud test is not the same as passing the paperwork test, and Supermicro has effectively been running a governance repair job and a margin recovery at the same time.

The most recent overhang, the Justice Department's export-control case involving a co-founder and two other individuals, remains unresolved. Supermicro itself was not charged, and it placed the employees involved on leave, but the board's own review is ongoing, and the July 21 update itself repeats a warning that this review "could affect our forecasts, these preliminary results and prior period results." That single sentence may matter more than the headline margin number, since it means even this update carries an asterisk.

The Trust Ladder: Has Supermicro Rebuilt Investor Trust After Its Accounting Crisis?

Here's a simple way to think about where Supermicro stands, something we'll call the Trust Ladder. Rebuilding investor confidence after a governance scare like this tends to happen in stages, not all at once.

On this ladder, Supermicro looks like a company past the emergency stage but still a couple of steps from being the kind of dependable business that commands a premium multiple. The July 21 update reads like an attempt to climb steps three and four simultaneously. Whether it succeeds shows up in the August 11 numbers, not in a preliminary press release.

What Are Analysts’ Latest SMCI Stock Price Targets?

Pooled together, aggregator data from MarketBeat and TipRanks around this announcement showed roughly 4 to 5 Buy ratings, 9 to 12 Hold or Neutral ratings, and 1 to 2 Sell ratings on the stock, with an average 12-month price target clustering between about $37 and $39.

A few things jump out. This is overwhelmingly a Hold-rated stock on Wall Street, not a consensus Buy, despite the AI story attached to it. Goldman Sachs is the most visibly bearish name of the group, and it reiterated its Sell rating specifically in reaction to the July 21 update rather than walking it back, which tells you at least one major desk remains unconvinced the margin jump changes the underlying risk picture. And nearly every target in that table predates the July 21 news itself (Goldman's is the only confirmed post-announcement reaction we could verify), so expect a round of target revisions once audited numbers land on August 11.

Is SMCI Stock a Value Play or a Growth Story?

This is really the question every reader wants answered, so let's tackle it through two different lenses rather than one verdict.

The Value Lens looks backward and asks whether the stock is cheap relative to what the business is already worth today. On this measure, Supermicro does look statistically inexpensive. At a market capitalization of roughly $16.5 billion against fiscal 2026 revenue guidance of nearly $40 billion, the stock trades under 1 times sales, with a forward price-to-earnings ratio somewhere in the high single digits to low double digits depending on which day's price you use, a low multiple for a company still growing revenue quickly. A value-minded investor would note the stock trades well below its own history and below most AI-adjacent hardware peers.

The Growth Lens looks forward and asks a different question: is the trajectory strong enough, and the balance sheet clean enough, to justify paying up for future earnings? Here the picture is more mixed. The $60 billion backlog and new margin guidance both support a genuine growth story if they hold up. But growth investors typically also want confidence in execution and clean books, and Supermicro is still mid-repair on both fronts, carrying more debt and still absorbing the dilution from that raise into its share count.

Put simply, the value case rests on Supermicro being statistically cheap for what it already does today. The growth case rests on this quarter's numbers being the start of a trend rather than a one-off. Both can be true at once. Both can also be wrong at once, if the DOJ matter drags on or margins snap back down. Neither lens alone gives a complete answer.

How Much Could Supermicro’s Margin Recovery Add to Gross Profit?

Here is a simple way to size the margin jump's dollar impact, using only the company's own guidance midpoints. If fiscal 2026 revenue lands near the guided $39.65 billion, moving gross margin from 8.3% to 16% would add roughly $3 billion of extra gross profit for the year, without selling a single additional server. But that is assuming that the new gross margin figure applies to the full year revenue and not just Q4. That is the kind of shift that can move a stock double digits in an evening, because it rewrites the profit story without touching the revenue line at all.

The other useful exercise is reverse-engineering what it would take to reclaim the old highs. At a price near $30, Supermicro would need to roughly quadruple from here to reach its split-adjusted peak of $122.90, assuming the market keeps valuing the stock at a similar earnings multiple to today. That is simple arithmetic: the price ratio between $30 and $122.90 works out to about 4.1 times, and if the multiple investors are willing to pay stays roughly constant, earnings need to grow by that same 4.1 times to get the stock back there. If the market were willing to pay a richer multiple than today, on the theory that the growth story is now fully proven, the earnings bar would be lower. Either way, this looks like a multi-year climb, not one good quarter.

None of this is a forecast. It is the arithmetic of what would have to be true for the bull case, or a genuine return to the old highs, using the company's own disclosed numbers as the starting point instead of an outside guess.

Our Take, And What We're Watching Next

The July 21 update is real news, not noise, but it answers a narrower question than the stock's price action suggests.

It tells us Supermicro received an enormous amount of new business and believes its margin profile is improving. It does not yet tell us whether that improvement is a durable shift in the business or a favorable one-quarter accident, because the company has not made that distinction itself, and because these are unaudited numbers from a company with a real, well-documented history of getting its own numbers wrong. The stock's 75% discount to its old highs is not purely an AI-hype hangover fading; a meaningful part of it is a trust deficit that one press release, however encouraging, cannot fully repair on its own.

For anyone tracking this story rather than acting on it, a few things matter more than the headline percentage moves: the August 11, 2026 earnings call, where audited figures and management's own explanation will either confirm or complicate this narrative; whether gross margin guidance for the following quarter stays anywhere near the mid-teens instead of reverting to single digits; any update on the DOJ export-control review the company keeps flagging in its own filings; and how much of the $60 billion backlog actually turns into recognized revenue over the next few quarters, since bookings and billings are not the same thing.