Microchip Technology Incorporated MCHP

Key data

The business

Microchip makes the small brains embedded in physical products: microcontrollers, the analog and power parts around them, memory, and programmable logic. A car holds dozens, a factory tool or a medical monitor a few each. The catalog runs to tens of thousands of parts sold to roughly 120,000 customers, no single one large, mostly through distributors. That breadth is the moat. Once a Microchip part is designed into a board, swapping it means re-engineering and re-qualifying the whole thing, so designs sit for years at gross margins above 60%. It is an integrated maker running its own older factories rather than chasing the leading edge, which keeps the plants cheap to feed. The same broad, distributed model made the 2024 downturn brutal: when end demand softened, distributors and customers who had double-ordered through the shortage slammed the brakes, and revenue fell by nearly half from its peak. The recovery now underway is the whole story, and everything below is a way of asking whether it is real.

The numbers

The shape of a cycle bottoming and turning is written across the last five quarters.

The trough was Q1 FY26. Since then revenue has climbed every quarter, and Q1 FY27 landed up 38% on the year and 13% sequentially, with net income swinging from a loss to $230M. Management calls the book-to-bill ratio well above one, the order signal that a recovery has legs. Diluted EPS is stated after the dividend on a mandatory convertible preferred, which is why the per-share line lags the profit line.

Peak to trough, revenue fell 48% from FY23's $8.44B to FY25's $4.40B, and free cash flow per share fell from $5.62 to $1.59. This is not a compounder read mid-stride. It is a cyclical caught at the bottom and turning up, and the multiple only makes sense if the turn carries to something near the old run-rate. What settles the argument is one print: whether quarterly revenue re-crosses the roughly $1.7B it ran at before the bust, and whether the margin comes with it.

The recovery shows cleanest in the margin, where fixed costs spread back over rising volume.

Ten points of gross margin and twenty of operating margin in a year is the tell of a real volume recovery, not cost cutting. The old peak sat near 68% gross and 37% operating, so there is room left if the top line keeps climbing.

Management

Steve Sanghi, the founder, took the CEO seat back to run the turnaround and was paid $40.5M in fiscal 2026, about 20% of the year's net income, nearly all of it a one-time equity grant tied to his return, the kind of number a board justifies by results not by precedent. Insider open-market activity is thin and one-sided: the CFO sold $377K across two August sales, small enough to read as housekeeping, with no offsetting buys. The buyback record is honest by restraint, repurchases cut from $982M in FY24 to $96M in FY25 to protect the balance sheet through the downturn. The dividend is the pressure point. In fiscal 2026 the $984M paid out still exceeded the $871M of free cash the business generated, so the shortfall leaned on borrowing much as it had the year before, but the recovery has since pulled trailing coverage back to roughly breakeven, so that watch-item is closing rather than closed. The 7.50% mandatory convertible preferred, issued to shore up the balance sheet at the bottom, costs about $28M a quarter and will convert into stock, dilution already moving through the count.

How it fails or surprises you

The recovery double-dips. If book-to-bill slips back under one and sequential revenue flattens near $1.5B, the whole recovery re-rates at once. Net debt at three times EBITDA and a dividend only just covered turn from tolerable to urgent, because there is no cost left to cut. Watch the next two book-to-bill readings against sequential revenue.

The multiple is paying for peak earnings not yet earned. At about 22 times forward earnings, you pay a peak-history price two-thirds of the way up from the trough. If the FY23 shortage inflated the old peak and normal earnings power sits lower, then mid-range multiple is the wrong frame and the stock is expensive. The print that decides it is whether revenue clears the old $1.7B quarter.

The full-cycle snapback (right tail). Embedded down-cycles overcorrect, and the up-swing has historically run harder than a mere recovery-to-normal. If gross margin re-crosses 65% and revenue passes the old peak inside four quarters, adjusted earnings could clear $4, a level the stock has fetched 20 to 24 times before. Today's price pays for none of that. The first tell is gross margin above 65%.

Closing thoughts

This is an uncertainty a named print resolves, not an exposure with no answer, and the distribution is bimodal around one question: return to the old peak, or return to a lower normal. My read is the fatter tail is the upside, because a margin recovery this fast and clean is what a real volume turn looks like and not what a stalling business produces, and embedded semis have snapped back hard from every prior inventory bust. The left tail is real all the same. Net debt at three times EBITDA and a dividend the business only just covers means a stalled recovery costs more here than at a debt-free peer. What is at risk if orders air-pocket again is a balance sheet with no more room to give.

The bet is still that factories, cars, and machines go back to buying Microchip's control chips at the old pace, the warehouses having drained. What breaks it is a second stall in orders before the balance sheet heals, and the one pair of numbers that tells you first is next quarter's book-to-bill against sequential revenue: both up and the bet is working, book-to-bill under one with revenue flat and it is not.