Roku finally turned a GAAP profit — but stock comp still dwarfs it, and the boxes still sell at a loss

---

For most of the past five years, the bear case on Roku wrote itself. A streaming pioneer with a commanding installed base and not a dollar of GAAP profit to show for it; a hardware division bleeding cash by design; an advertising engine that grew impressively but never quite escaped the gravity of its own operating expenses. Skeptics could simply point at the bottom line and rest their case. That line of attack is now closed. In February 2026, Roku reported its first full year of GAAP net income since the pandemic boom of 2021 — $88.4 million on $4.74 billion of revenue, a swing of more than $200 million from the $129 million loss it booked in 2024. Then, on April 30, the first quarter of 2026 landed even better than the bulls dared hope: revenue up 22% year-over-year to $1.248 billion, net income of $85.7 million versus a $27.4 million loss a year earlier, and adjusted EBITDA up 165% to $148.4 million. The stock popped. The headlines wrote themselves the other way this time.

So let us be precise about what we are arguing, because the lazy short thesis — "Roku doesn't make money" — is now factually dead, and any analyst still repeating it is trading on stale information. The interesting question is not whether Roku is profitable. It is whether the profit it has produced is the durable, self-funding, defensible kind that justifies the valuation the market has assigned it — or a thin, fragile, dilution-subsidized sliver of black ink that the company has reached by running every other lever to its limit at exactly the moment the largest companies on earth decided to compete for its core revenue. The numbers Roku itself reported, read carefully, point toward the second reading. This is a forensic walk through why.

The profit is real — and it is smaller than the stock-comp bill

Start with the single most important relationship in Roku's income statement, the one the celebratory headlines stepped right over. In fiscal 2025, Roku produced $88.4 million of GAAP net income. In that same year, it recorded $354.2 million of stock-based compensation expense. The compensation paid to employees in stock — a real cost, a real transfer of ownership away from existing shareholders — was approximately four times the entire reported profit of the enterprise.

This is the central tension in the bull narrative, and it is not a rhetorical trick. Stock-based compensation is a genuine economic expense; the accounting bodies that require it on the income statement are right to do so. When a company pays its workforce in shares, it is handing over slices of the business that would otherwise belong to outside owners, and the only reason this does not show up as cash leaving the building is that the dilution is borne silently by the share count instead. Roku's GAAP net income already deducts that $354 million — which is precisely why the profit is so thin. The company has reached profitability not by overwhelming its SBC line with operating leverage, but by clearing it with barely $88 million to spare. Strip out a single bad quarter, a single impairment, a single missed ad season, and that cushion vanishes.

Management deserves credit for moving in the right direction here, and we will return to that in the bull section. SBC fell 8.1% year-over-year, from $384.7 million in 2024 to $354.2 million in 2025, and is guided to roughly $325 million in 2026, with the company explicitly claiming "a clear path to fully offset dilution in 2026." That is the correct goal and a meaningful improvement. But note what the framing concedes: as of today, dilution is not yet fully offset, and the cost of the workforce in stock still exceeds the company's GAAP earnings by a multiple. The bull who tells you Roku "earns" must reckon with the fact that the company earns less in a year than it pays its own people in equity. That is not a fraud. It is a quality-of-earnings problem, and it is the difference between a business that prints cash for owners and one that has merely stopped destroying it.

The boxes that build the empire are sold at a deepening loss

Now to the hardware — the part of the story that has not improved, that in fact got worse in the very quarter the bulls are celebrating. Roku's Devices segment carries a negative gross margin. Not a thin margin, not a low margin: a negative one. In Q1 2026, devices gross margin deteriorated 250 basis points year-over-year to negative 16.3%. Devices revenue fell 15.9% to $117.6 million, dragged down by lower streaming-player unit sales and promotional pricing, with the company itself citing persistent average-selling-price pressure and higher memory costs.

This is not an accident or a temporary stumble. It is the deliberate architecture of the business. Roku has run its hardware arm as an intentional loss leader since 2021, selling players and licensing the Roku TV operating system at or below cost to expand the installed base. Negative gross margins on devices averaged roughly negative 12.1% across 2023 to 2025; for full-year 2026, management guides devices revenue to grow only low-single-digits to about $610 million, at a gross margin in the "negative mid-teens." Read that again: the company is forecasting that it will continue, on purpose, to lose money on every box it ships, for the entire coming year, in order to recruit the households that the platform business then monetizes.

The bull frame for this is "customer acquisition cost," and it is a legitimate frame — a household subsidized for a few dollars at the point of sale can throw off platform revenue for years. But the forensic observation is sharper than the bull frame admits. The hardware subsidy is now deepening at the very moment hardware unit sales are falling. You are paying more, per unit, to acquire fewer units. That is the precise signature of an acquisition channel under competitive pressure: to keep the funnel full, you have to discount harder, and even then the volume shrinks. The boxes are getting more expensive to give away and the giveaway is recruiting fewer people. A loss leader that is losing more to lead fewer is a loss leader whose economics are inverting — and Roku has guided that inversion to persist all the way through 2026.

Where the money actually comes from — and who is coming for it

If devices lose money by design, the entire enterprise rests on a single load-bearing column: the Platform segment, which sells advertising and takes a cut of subscription sign-ups against that installed base. In Q1 2026, platform revenue was $1.13 billion, up 28% year-over-year, of which advertising was $613 million (up 27%) and subscriptions were $519 million (up 30%). Hardware is now less than 10% of total revenue. This is the whole business. The thesis stands or falls on whether the high-margin ad and subscription dollars keep compounding fast enough to lever past that $300-plus-million stock-comp bill and produce a profit that grows rather than merely exists.

And here is the part the milestone headlines elide entirely: the connected-TV advertising market that Roku pioneered is no longer Roku's to dominate. The three largest forces in digital advertising and streaming have all moved into the exact inventory Roku depends on. Amazon turned on advertising across Prime Video by default, instantly creating one of the largest premium CTV ad pools in existence, sold by a salesforce that already controls retail-media measurement. Netflix, which spent a decade swearing off advertising, now runs a fast-scaling ad tier across its own enormous engaged base — a base that watches longer and is harder to multi-home than a Roku home screen. Google sells YouTube CTV, the single most-watched streaming surface on American televisions, through the most sophisticated ad-buying machine ever built. Each of these competitors has a structural advantage Roku cannot replicate: first-party purchase data, proprietary premium content, or both. Roku's advantage was the operating system on the glass — the home screen, the remote, the ad real estate around the content. That is a real and valuable position. But it is a tollbooth on a road that three trillion-dollar companies are now building bypasses around.

The denominator that flatters every metric

Watch how Roku's favorite numbers are constructed, because the framing does a lot of quiet work. The company crossed 100 million streaming households on the Roku TV operating system and reported 38.7 billion streaming hours in a single quarter, up 8% year-over-year. These are genuinely large, genuinely impressive figures, and they are exactly the figures designed to keep your eye off the income statement.

Here is the forensic move. Households and hours are the denominator of Roku's growth story, and they are growing far slower than the revenue Roku extracts from them. Streaming hours grew 8% year-over-year while platform revenue grew 28%. That gap is the entire bull case stated honestly: Roku is monetizing each existing hour and each existing household much harder than it is adding new ones. Monetization-per-user growth is a perfectly good thing — until you ask how much further it can run, and against whom. The reach is maturing. Domestic household growth is slowing as the addressable base of cord-cutting US homes fills in; the 8% hours growth is a fraction of the platform revenue growth precisely because the user base is no longer the engine. The engine is price — the rate Roku charges advertisers per impression and the take it keeps on subscriptions. And price is the one variable that competition attacks most directly. When Amazon and Netflix flood the market with premium CTV impressions, the per-impression rate that lets Roku grow revenue at 28% on 8% more hours is exactly what gets compressed. The denominator is maturing while the numerator is being contested. That is the worst possible configuration for a business priced on the assumption that both keep climbing.

Adjusted EBITDA up 165% — net income up far less

There is a tell in how Roku leads its own earnings narrative, and quality-of-earnings analysis lives in tells. The Q1 2026 release trumpeted adjusted EBITDA of $148.4 million, up a spectacular 165% year-over-year. That is the number the company wants in the headline. GAAP net income was $85.7 million. The gap between those two figures — roughly $63 million in a single quarter — is the stuff that adjusted EBITDA adds back: stock-based compensation, depreciation, amortization, and the rest of the non-cash and "one-time" line items.

The 165% growth rate is real arithmetic, but it is flattered by a low base — adjusted EBITDA was small a year ago, so a given dollar increase looks enormous in percentage terms. More importantly, the metric Roku chooses to lead with is the one that adds back the single largest cost discussed above: the stock compensation that exceeds the company's actual profit fourfold. When a company has just achieved GAAP profitability after five years of losses — a genuine milestone it has every right to celebrate — and still chooses to put adjusted EBITDA in the headline above net income, that choice tells you which number it believes is more flattering. The forensic reader's job is to notice when the more flattering number is the one being waved, and to anchor on the more conservative one. On the conservative measure, Roku earns about $86 million a quarter, against a stock-comp run-rate alone of more than $80 million a quarter. The celebration is warranted. The proportion is sobering.

Priced for the contest to already be over

Roku has raised its full-year 2026 outlook to roughly $5.5 billion of revenue, with net income guided to about $325 million and adjusted EBITDA to about $635 million, implying margin expansion to 11.6% and more than 50% adjusted-EBITDA growth. Take that guidance at face value — and the company has just demonstrated it can beat its own numbers. Even then, the asymmetry is unfavorable for anyone buying at the current valuation.

Here is why. A stock that has just turned its first profit, trades at a premium multiple, and guides to triple-digit-millions of net income is being priced for a smooth, multi-year ramp from thin profitability to fat profitability. That ramp requires three things to all go right simultaneously: ad rates hold despite Amazon, Netflix, and Google flooding CTV supply; the hardware subsidy stops deepening even though management just guided it to keep deepening through 2026; and stock-based compensation continues to fall fast enough to actually offset dilution rather than merely slow it. Each of those is plausible on its own. The market is pricing all three as near-certain, together. That is the definition of priced-for-perfection: the good case is in the stock, and the merely-okay case — ad rates compress a few points, hardware stays a drag, SBC plateaus — is not a disaster for the business but is a serious disappointment for the multiple. When the asymmetry is "modest upside if everything compounds, real downside if anything normalizes," the risk-reward favors patience, not the chase.

The moat is a home screen, and home screens get bypassed

The deepest question is whether Roku's advantage is a moat or a loophole — a durable structural barrier, or a temporary position that competitors are actively engineering around. Roku's genuine asset is incumbency on the television itself: tens of millions of TVs ship with its operating system, its home screen is the first thing a viewer sees, and the ad real estate around that screen is valuable precisely because it is unavoidable. For years that was a moat, because the alternative — Amazon's Fire TV, Google's TV platform — competed for the same glass without decisively winning it.

But a home-screen tollbooth is structurally vulnerable in a way a content moat is not. Netflix does not need Roku's home screen; viewers open the Netflix app directly, and Netflix sells its own ads inside its own content. YouTube is the same — it is the destination, not a tile someone scrolls past. As viewing consolidates into a handful of apps that users launch by habit, the strategic value of being the operating system underneath those apps erodes; the home screen becomes a launcher people bypass rather than a marketplace they shop. Roku's response has been to push its own free ad-supported channel and its own content tiles, which is rational, but it puts Roku in direct content competition with companies that spend more on a single tentpole series than Roku earns in a year. The moat-versus-loophole test asks: if the competitor simply ignored your barrier, would your economics survive? For a content owner, ignoring you is impossible — they need your pipes. For an operating-system tollbooth, ignoring you is exactly what an app-first viewing world does by default.

What the bulls genuinely get right

It would be dishonest to end without conceding, specifically and fairly, that the bull case on Roku is stronger today than at any point in five years — and on several axes it is simply correct.

First, the profitability is real and it is improving, not deteriorating. This is not a company torturing its accounting to manufacture a one-time gain. Roku swung from a $129 million GAAP loss in 2024 to $88.4 million of GAAP net income in 2025, and then to $85.7 million in a single quarter in Q1 2026 — a genuine, accelerating reversal, confirmed in SEC filings, not a press-release adjustment. The trajectory is the right direction at the right slope.

Second, management is attacking the exact criticism leveled here. The stock-based compensation that this article dwells on is falling — down 8.1% in 2025, guided lower again in 2026 — and the company has stated an explicit goal to fully offset dilution this year. That is precisely the discipline a skeptic should demand, and Roku is delivering it rather than dismissing it. A bear who ignores that the company is already fixing the headline weakness is not being honest.

Third, the platform business is compounding at a genuinely impressive rate — 28% platform revenue growth, 27% advertising, 30% subscriptions — against a maturing user base, which means Roku has demonstrated real pricing power and real monetization depth, not just reach. Crossing 100 million households on the operating system is a scale that few competitors can match on US televisions specifically, and incumbency on the glass is worth something durable even in an app-first world. The loss-leader hardware strategy, for all its deepening drag, has actually worked: it built the installed base that now throws off high-margin platform dollars. And full-year 2026 guidance to $5.5 billion of revenue, $325 million of net income, and $635 million of adjusted EBITDA — from a management team that just beat its own numbers — is not a fantasy. If the CTV ad market grows the whole pie fast enough, there may be room for Amazon, Netflix, Google, and Roku to all win. The bull case is not that the competition isn't coming. It is that the market is large enough, and Roku's position strong enough, that the toll still gets collected.

The kicker

So the milestone is real, and the milestone is thin. Roku has done the hard thing — it has stopped losing money, and it is doing it with rising discipline on the one cost that mattered most. The risk was never that the company is a fraud or a zombie; it plainly is neither. The risk is one of proportion and timing: a company that earns less in a year than it pays its workforce in stock, that loses a deepening margin on every box it ships to recruit a slowing stream of new households, that depends entirely on advertising rates the three richest companies in technology have just decided to compete for — and that is priced as though all of those pressures resolve in its favor at once. That is not a short on the business. It is a short on the certainty the market has assigned to the happy ending.

The question is not whether Roku can make money — it just proved it can; the question is whether a profit smaller than the stock-comp bill, financed by boxes sold at a loss and ad rates three trillion-dollar rivals are now bidding away, is the durable kind the market has already paid full price for.