MACOM's AI Optical Boom Is Real — That's What Makes the Price Hard

MACOM Technology Solutions reported a quarter that looks almost impossible to argue with. In the three months ended July 3, 2026, revenue rose 35.8% from a year earlier to $342.2 million, adjusted earnings hit $1.40 a share, and management promised roughly $415–425 million in the next quarter — another 21% to 24% sequential jump. The driver was named on the call: data-center revenue reached $137.6 million, up about 40% sequentially, on demand for the high-speed optical parts that move data inside AI clusters. The book-to-bill hit a record 1.6 to 1, meaning orders are still piling in faster than they are shipped.

That is the explosion. The very big question behind it is simpler than the numbers: can the market's price for MACOM outlast the pace that produced it?

Start with what the price already assumes. Even after the stock sits about 36% below its 52-week high of $418, it trades at roughly 17 times trailing sales and about 85 times trailing earnings, for a market cap near $20.5 billion. That multiple is not an outlier in this corner of the market — Marvell, Lumentum, and Broadcom are all priced at 19 to 26 times sales. But it means MACOM is being valued as a durable, multi-year AI-optics winner, not as a company selling into a hot quarter.

The tension is one of timing. The growth powering today's number is concentrated in 800-gigabit and 1.6-terabit PAM4 components — the current generation of AI optical interconnect. Total revenue grew 18% in a single quarter, and the data-center piece grew about 40% in that same stretch. Growth like that cannot be extrapolated; it is the steepest part of a ramp that must eventually flatten. The question is where the plateau sits and whether anyone sees it coming.

Management's own roadmap shows how far out the next legs of the story are. The follow-on products that would extend this wave — 75-milliwatt CW lasers and near-packaged optics for next-generation scale-up fabrics — are not expected to contribute meaningful revenue until late 2027 and 2028, respectively. That leaves the next several quarters resting on the current PAM4 generation plus hyperscalers' continuing appetite for AI data-center buildout, a capital-spend cycle known for lumpiness and inventory corrections. The record booking pace cuts both ways: it gives near-term cover, but today's pulled-in orders are part of how a hot cycle turns into a correction when demand steps down.

None of this is a broken company. MACOM's economics are genuinely improving, with adjusted gross margin at 59.7%, adjusted operating margin at 31.5%, and a comfortable balance sheet: roughly $663 million in cash and short-term investments against convertible notes, for a net cash position around $322 million. The company has also been deliberately widening its base — more than 20 customers now generate over $10 million in annual revenue, up from eight two years ago, and the largest ten end-customers are under 40% of year-to-date revenue. That is real execution against real demand, and it deserves to be distinguished from the valuation question.

The distinction is the whole analysis. MACOM is a good company with an improving cost profile riding a genuine, well-evidenced AI-optics boom. Whether it is a good stock at this price depends on a belief that hyperscaler AI spending keeps compounding at something close to its current pace long enough for the current product wave to carry into the next one. The multiple has already paid for much of that belief; it leaves little room for the inevitable quarter where double-digit sequential growth starts to cool.

That is the live test, and it is on a quarterly clock. Watch the data-center optical number and the book-to-bill on each print — a reversal there is the signal that the market has priced the continuation while the operating reality slows. Until the current wave either keeps compounding or hands off cleanly to the 2027–2028 products, the honest read is not to reach for a rating but to respect what the price already demands. The earnings explosion is real. The question the market is really asking is whether it can keep paying for it.