Harmonic's Broadband Pivot: The Pullback, The Proof, And The Story The Market Is Still Missing

Harmonic's Broadband Pivot: The Pullback, The Proof, And The Story The Market Is Still Missing

Harmonic reported its best broadband quarter to date, beat earnings, and raised full-year guidance. The stock fell 10% over the next five days.

The GAAP headline — a quarterly net loss of $0.02 per share — explains the reflexive sell-off. But it also explains almost nothing about the business Harmonic actually operates now. That loss came from its former video division, which was sold in mid-June. The continuing broadband business posted $31.3 million of operating profit and 53% gross margin.

The disconnect between the tape and the operating reality is the point. Harmonic has completed a structural transformation that changed its revenue growth, its balance sheet, and its earnings trajectory. The market has yet to let the old narrative fully go.

The old story is gone. The proof is in the numbers.

Harmonic was once a combined broadband-and-video company. The video segment dragged on growth and margins while competing in a crowded, commoditized market. In December 2025, management announced the sale of the Video business to MediaKind for $145 million in cash. The deal closed on June 16, 2026.

That was the structural change. Now the company operates a single reportable segment: virtualized broadband. The question wasn't whether the change was happening — it was whether the broadband business that remained could grow.

The second quarter of 2026 answers that question.

Broadband revenue grew 54% year over year to $133.5 million. That wasn't a one-off spike — Q1 2026 broadband revenue was up 43%, and Q2 accelerated past the high end of guidance ($115-$125 million). The rest-of-market segment — customers outside the two largest — grew 44% to $49.4 million and now represents 37% of revenue. More importantly, rest-of-market bookings reached 60% of the total in Q2, with a book-to-bill ratio well above 1.5.

Backlog and deferred revenue hit $587.6 million, up 71% from a year earlier. The cOS virtualized platform is commercially deployed with 161 customers, serving 48.2 million cable modem endpoints. Operators are using it to transition from DOCSIS upgrades to fiber without ripping out existing infrastructure, and new products like the Pearl-1XL and Oyster+ fiber solutions have already secured multimillion-dollar orders.

This is not a company hoping for a turnaround. It is a company with a defined product, a growing installed base, and a backlog that's nearly doubled in a year.

The quarterly loss is a one-time accounting event. Cash tells a different story.

The GAAP net loss of $0.02 per share came from the video divestiture — $19.4 million in discontinued operations charges, including a $6.3 million loss on the sale and stranded costs. Continuing operations generated $17.1 million in GAAP net income. Non-GAAP EPS was $0.24, beating consensus of $0.17 by $0.07.

Management flagged about $10 million in total stranded costs across the full year from the divestiture, with a target to eliminate 30% within 12 months. There's also an ongoing headwind of roughly $3 million per quarter in unrecovered memory component costs that can't be passed to customers. These are real cost pressures, and they matter. But they're known quantities, and management has already baked them into the raised guidance.

The cash flow picture is where the story becomes concrete.

Free cash flow over the trailing twelve months sits at roughly $54 million — generated from a business that produced $62 million in operating cash flow with only $8.4 million in capex. The balance sheet is clean: $232 million in cash against $272 million in debt, for a net cash position of about $121 million. Current ratio is 305%. Inventory grew to $66.5 million as the company secured memory supply for anticipated demand — a proactive cost, not a sign of weak orders.

This is not about excitement. It is about a business that may soon look a lot harder to dismiss once the free cash flow from the transformed, broadband-only operation shows up as the standard TTM metric.

What the valuation is actually saying

Harmonic's market cap is $1.31 billion. At the current price around $12, the stock has retreated from its 52-week high of $17.68 — a decline of roughly 32%.

The published forward P/E of 74x tells you more about stale consensus than about the business. That multiple is built on consensus estimates that still reflect the old, slower growth profile. Management's own guidance paints a different picture.

For the full year 2026, Harmonic is guiding to $0.67-$0.75 in non-GAAP EPS, with $0.44-$0.53 on a GAAP basis that includes the stranded costs. At $12, the midpoint of $0.71 EPS implies a forward multiple of about 17x. The GAAP midpoint of $0.485 implies roughly 25x. Either way, these are forward-looking ranges for a business that just grew revenue 54% and raised guidance at the high end.

The stock traded at a significantly higher multiple at the 52-week high, before the guidance raise and the video divestiture was complete. The pullback to $12 compresses that multiple into a range where a 50%+ growth rate actually looks like value rather than premium pricing.

The market is still pricing the old risk profile — video baggage, customer concentration, margin erosion — while the operating setup is already cleaner.

What goes wrong

The bear case is straightforward, and it deserves candor rather than dismissal.

Fifty-four percent growth comes off a year when the video business was still a drag. If broadband revenue growth decelerates sharply in Q3 and Q4, the raised guidance of $505-$525 million could come apart. Memory costs are adding roughly $12 million annually in unrecovered expenses, and the $10 million in stranded costs from the divestiture will eat into margins through year-end. Management is guiding to 51-52% gross margin for the year; if that drops, the earnings path flattens.

Customer concentration still matters. Even though rest-of-market bookings reached 60%, revenue from those customers is only 37% of the total. If one of the two largest customers reduces spending, the impact is material.

And the backlog, while impressive at $587 million, includes deferred revenue on work already performed — it's not a guaranteed revenue lock-up for future periods.

The thesis breaks if broadband revenue growth slows below the pace that supports the $0.67-$0.75 EPS guidance, or if memory cost inflation accelerates beyond the $3 million per quarter already priced in. Either scenario would mean the transformation is working less cleanly than management expects.

The proof path

The setup is simple enough to evaluate. Harmonic will report Q3 2026 results in early November, guiding to $125-$135 million in broadband revenue. The full-year midpoint is $515 million. Against a run rate of $266 million in the first half (Q1 $121.7M + Q2 $133.5M), the second half needs to hit roughly $249 million. That's achievable — it works out to about $124 million per quarter, right at the midpoint of Q3 guidance.

The non-GAAP EPS of $0.67-$0.75 has already been beaten in Q1 ($0.21) and Q2 ($0.24), leaving $0.22-$0.30 for the second half. The book-to-bill ratio is above 1.5, the backlog is growing, and rest-of-market demand is accelerating. If management delivers on its own guidance, the forward multiple at $12 stays in the 16-18x range for a company growing revenue in the 40-50% range with 50%+ gross margins.

The earnings path doesn't require complex DCF models or macro assumptions. It requires broadband revenue to grow, margins to hold, and the stranded costs from the divestiture to fade through the next year.

A beaten-down multiple on its own is not evidence — it can reflect permanent deterioration. But here the multiple compresses alongside accelerating revenue, a growing backlog, a cleaner balance sheet, and raised guidance. The market is pricing the old story while the numbers already point to the new one.

I can be wrong again. But the setup is concrete: broadband revenue grows 54%, the balance sheet has $232 million in cash against $272 million in debt, the backlog doubled, and management raised guidance into the fourth quarter. The break condition is equally clear — if broadband growth decelerates below the pace needed to hit the midpoint of the raised guidance, the rerating case doesn't hold.