Watch the full analysis on YouTube!

On May 5, 2026, Supermicro reported a quarter that tells you almost everything you need to know about this business. Revenue came in at $10.2 billion — huge by any measure, but down from $12.7 billion the quarter before. Net income rose to $483 million. And yet the company burned $6.6 billion of cash in operations in a single quarter. Read that again. A company posting record sales drained $6.6 billion of cash in three months.

That gap — between the headline numbers and the cash — is the whole story. Supermicro is one of the hottest names of the AI boom. It builds the servers that house Nvidia's chips, and demand is real. But this is not a new company. It was founded in 1993 and has traded publicly since 2007. Nineteen years in, it still runs razor-thin margins, it still can't produce consistent free cash flow, and it has a track record with regulators that should make any serious investor slow down.

Here is my honest starting point. I love a business that trades at or below fair value. But what I've learned to love more is the quality of the business. Price is what you pay; quality is what you actually own. So in this analysis we're going to do two things together: figure out what Supermicro is worth, and — just as important — figure out whether it's the kind of business you'd want to own at any price.

In plain English, Supermicro builds computer servers. A server is the powerful computer that lives in a data center and does the heavy lifting behind everything online — websites, cloud software, and now artificial intelligence. Supermicro designs the building blocks (motherboards, chassis, power, cooling) and assembles them into complete systems tuned for a customer's exact workload. Today the prize is AI servers: the racks that hold Nvidia's GPUs and the liquid-cooling systems that keep them from melting.

How this industry actually works. Think of Supermicro as a master tailor, not the mill that makes the cloth. The most expensive part of an AI server — the GPU chip — is bought from Nvidia. Supermicro adds engineering, integration, and speed-to-market around it. That's valuable, but it means the company sits in the middle of the value chain. The chipmaker captures the fat margins; the assembler captures volume. When AI demand surges, Supermicro's revenue explodes — but so does the cost of the components it has to buy, which squeezes how much it actually keeps.

Why it's cyclical. Server and chip demand moves in waves. Hyperscalers and enterprises spend heavily to build out capacity, then pause to digest it. That boom-and-bust rhythm shows up directly in Supermicro's numbers — explosive growth one year, air pockets the next. Right now we are mid-boom, which is exactly when a business looks its most impressive and is easiest to overpay for.

The two segments. Supermicro reports in two buckets. Server and Storage Systems is essentially the entire company — $21.3 billion of FY2025 revenue, 97% of the total. Subsystems and accessories (components sold on their own) is just $660 million, about 3%, and it's actually shrinking. So when we evaluate Supermicro, we are really evaluating one thing: its ability to assemble and sell AI and enterprise servers profitably.

Where they stand in the industry. Supermicro is a real player in AI servers — a fast-growing challenger that has won meaningful share — but it is not the king of its industry. It competes with far larger, better-capitalized rivals like Dell and Hewlett Packard Enterprise, plus the Taiwanese contract manufacturers. As you'll see in the margin and value-creation numbers below, Supermicro consistently operates below the average of its own industry. That's the tell. A company can be growing fast and still be one of the weaker operators in its field.

Start with the good news, because it's genuinely impressive. Revenue went from $2.2 billion in FY2016 to $21.9 billion in FY2025, and the trailing-twelve-month figure is already $33.7 billion. Over the past five years revenue has compounded at roughly 49% a year. On the surface, that's a rocket ship.

Now turn the numbers over. As revenue exploded, the margins went the wrong way. Gross margin — the slice left after paying for the parts — was about 15% for years. As the AI-server mix took over, it fell to 13.8% in FY2024 and just 11.1% in FY2025, with recent quarters dipping into single digits (6.3% in Q2'26 before recovering to 9.9% in Q3'26). Net margin tells the same story, sliding to 4.8% in FY2025.

Why margins shrink as revenue grows. This is the heart of the cyclical-assembler problem. The newest AI servers cost dramatically more to build — premium GPUs, exotic cooling, complex integration. Supermicro is selling far more, but each dollar of sales now carries more bought-in cost. So you get the strange picture of a company getting bigger and less efficient at the same time. When I look for quality, I want to see margins hold or expand as a company scales. Here they're doing the opposite.

Revenue and margin history

Here's the line that should stop you. Supermicro's five-year average gross margin is 14.6%. The computer-hardware industry average is 38.4%. Net margin: Supermicro 6.0% versus an industry average of 17.8%. Right off the bat, that tells me I should probably be looking at one of its competitors instead. And remember — this isn't a scrappy startup that hasn't found its feet. This is a 19-year-old public company. The thin margins aren't a phase; they're the model.

Free cash flow — the metric I trust most.

Earnings can be massaged. Free cash flow — the real cash left after running and reinvesting in the business — is much harder to fake. So this table matters more than any other in the report.

*FY2022–FY2021 operating cash flow and CapEx shown approximately; free cash flow is exact. Source: Wisesheets / SMCI 10-K.

Look at that chart. Over the last ten years, Supermicro had four years of negative free cash flow. This company profits on paper, but through its operations it often doesn't bring in cash — it consumes it to fund inventory and receivables. And the most recent data is the loudest warning of all: in the first nine months of FY2026, operating cash flow was negative $7.6 billion as the company poured cash into a $11.1 billion inventory pile and $8.4 billion of receivables to chase AI orders.

A note on normalized free cash flow. FY2025's positive $1.53 billion is the best recent year, helped by receivables coming down. It is the figure we'll carry into the valuation as the base. But be clear-eyed: given the swing back to massive cash burn in FY2026, even that $1.53 billion may flatter the true, through-the-cycle cash generation of this business.

Debt to free cash flow (the primary test). This answers a simple question: how many years of free cash flow would it take to pay off the debt? My benchmark is 3× or less = great, 5× or less = solid, above 5× = concern. At FY2025 year-end, long-term debt of $4.7 billion against $1.53 billion of normalized FCF was about 3.1× — solid. But the picture is deteriorating in real time. By March 31, 2026, total bank debt and convertible notes had ballooned to $8.8 billion. Against the same $1.53 billion FCF base, that's roughly 5.7× — into concern territory — and that's before accounting for the fact that current cash flow is deeply negative.

Debt to equity. At June 2025 the company actually held more cash than debt — a net-cash position. Just nine months later it carries net debt of about $7.5 billion, with debt-to-equity climbing from roughly 0.75× to about 1.16×. The direction is clearly the wrong way.

Excess cash and flexibility. Cash fell from $5.2 billion in June 2025 to $1.3 billion by March 2026 — drained by the working-capital build. That is a meaningful loss of financial flexibility right when the company says it needs to fund even more growth. Interest coverage is still adequate — operating income covered interest about 10–11× over the past nine months — but interest expense is rising quickly as the debt stack grows.

Supermicro pays no dividend. Zero. The company keeps every dollar to reinvest in growth (and, lately, to plug a cash-flow hole). For an income investor looking for a check in the mail, this stock does nothing for you. For a growth investor, the absence of a dividend is fine if the reinvested cash earns a strong return — but as the Value Creation Test below shows, the returns here are unremarkable for the industry.

Which sleeve does it belong in? With no dividend and a cyclical, high-growth profile, SMCI is strictly a growth-sleeve candidate — never a holding for the 65% value/income side of a portfolio. And as a growth holding, it has to clear a quality bar it currently doesn't reach.

This is the section most investors skip, and it's exactly why Supermicro's recent news rattled the market. When a company issues more shares, every existing owner's slice of the pie gets smaller. Same business, more owners, less for you.

Share count has risen almost every single year since the 2007 IPO — from about 479 million in FY2016 to 594 million in FY2025, total dilution of roughly 24% over nine years (~3% a year), and accelerating (FY2024 alone added 6.8%). The driver is partly stock-based compensation — about $314 million in FY2025 and already $306 million in just nine months of FY2026 — and partly outright equity raises.

And it's about to get worse. The company has signaled plans to raise roughly $7 billion through equity and equity-linked instruments — convertible notes that can turn into stock. That's exactly what spooked investors. Add convertibles to the picture and the fully diluted share count is already about 695 million versus 594 million basic. Here's my problem with it: given a history of thin margins and inconsistent cash flow, why would I trust this company to pay off new debt rather than simply convert it to equity and dilute me further? Can't trust it. When you understand the trend, investors selling on that news look less like panic and more like a savvy move.

This is one of my favorite quality checks. It compares the return a company earns on its invested capital (ROIC/ROCE) against its cost of that capital (WACC). If the return is higher than the cost, the business creates value. If it's lower, it destroys value. The gap between them — the spread — is the prize.

At first glance, Supermicro passes: it earns about 17.8% on capital against a 10.5% cost, creating value at roughly +7.4% a year. That's not bad on its own. But quality is relative. The computer-hardware industry creates value at about +35% — nearly five times Supermicro's spread. Once again, the company sits well below the average of its own field.

A fair caveat: the industry comparison group includes hardware companies that aren't all direct server competitors, so treat it as a benchmark rather than a perfect apples-to-apples peer set. Even on a standalone basis, though, a +7.4% spread from a business this cyclical, this thinly profitable, and with this much governance baggage is not the profile of a premium compounder.

This table shows how management deployed its operating cash in FY2025 — what their real priorities are, straight from the cash flow statement (percentages are of $1,659M operating cash flow).

Two takeaways. First, nothing goes to shareholders as dividends, and only a modest $200 million went to buybacks — which is more than offset by ongoing dilution. Second, and more important, this snapshot understates reality. The bigger story is that Supermicro raised roughly $2.9 billion of convertible notes in FY2025 and is now consuming billions to fund inventory and receivables. Management's true capital priority isn't returning cash or even fixed assets — it's feeding the working-capital machine required to chase AI orders. That can work in a boom. It's brutal if demand pauses.

I use a discounted cash flow (DCF) model — projecting future free cash flow and discounting it back to today's dollars. Here are my inputs, and the reasoning behind each.

On the discount rate. The company's calculated cost of capital is about 10.5%. I'm deliberately using 15%. Why higher? Because of everything above — the thin margins, the inconsistent cash flow, the coming dilution, and the governance history. A lower-quality, higher-risk business should be made to clear a higher bar. That's my hurdle rate for this name.

On the terminal multiple. Three times free cash flow sounds brutally low — and it is. But look back at the stock chart: outside one brief window in 2024–2025, the market has rarely been excited about this company. Over its 19 years public, roughly 3× free cash flow is about where it has actually traded. I'd rather assume Supermicro reverts to its long-run multiple than bet it stays a market darling.

Projected free cash flow (44% growth)

How the model lands. Because Supermicro doesn't reliably throw off free cash to its owners along the way — it reinvests (and often burns) it — the value sits in what a buyer would pay for the business down the road. Discounting that exit value back at 15% produces an intrinsic value of about $21.54 per share, with a fair-value range of $19 to $24 (the low end reflects a 4-year exit, the high end a 5-year exit). The range is the real output here — valuation is an art, not a science.

Price vs. value. The stock trades at $30.46. That's about 29% above intrinsic value and sits clearly above the entire fair-value range.

Implied growth check. Working backwards from today's price, the market is effectively pricing in roughly 61% annual free-cash-flow growth — well above the 49% Supermicro has actually averaged and far above my 44% assumption. In other words, the market is betting this boom not only continues but accelerates. I'm not willing to make that bet at this price for this business.

At $30.46, Supermicro trades about 29% above its intrinsic value and well above the top of the $19–$24 fair-value range. More to the point, this isn't a premium business I'd happily own at a premium price. The margins run far below the industry, free cash flow is inconsistent and currently deeply negative, the value-creation spread is a fraction of peers, dilution is climbing with $7 billion more on the way, and the company carries a long shadow over its accounting and controls. When you're paying up for a below-average, higher-risk business in the most exciting part of its cycle, the math and the quality both point the same direction. The verdict is Sell.

_____________________________________________________________________________

This business fails my analysis due to the lack of quality and untrustworthy management.

It is clear that Supermicro’s hype is due to the AI boom and what they could be, but not who they are in reality.

_____________________________________________________________________________

This analysis is published by The Investing Department for educational purposes only. It is not financial advice. Always do your own research and consult a qualified financial advisor before making any investment decisions.