Netflix stops counting subscribers, starts selling the second act

On 16 June 2026, the trading day after Netflix reported its Q1 2025 numbers — $10.5B in revenue, $6.61 in earnings per share, a 25% beat on the bottom line — the market did something instructive. It barely moved. The same Wall Street Journal scoop that lifted shares in the hours after the print, a leaked internal plan to roughly double revenue to $78B by 2029 and reach a $1T market capitalisation, was treated as confirmation of a thesis investors had already paid up for. Netflix trades at roughly 50 times earnings. The script, the joke goes on the trading desks, is the product.

The disclosure that Netflix has stopped reporting quarterly net subscriber additions — for the first time in the company's public history — sits at the centre of that shift. It is a small change in disclosure that is large in meaning. The metric that built the streaming era, the quarterly add, has been retired at the very moment the underlying business crossed into the territory where growth is no longer the story.

A subscriber metric retires itself

The case for retiring the number, as the company framed it on the call and as analysts have parsed it, is straightforward: subscriber growth in lower-priced markets distorts the headline. A million new sign-ups in India at $2.50 a month is not the same business event as a million in the United States at $17.99. Netflix now serves more than 700M people globally, who together watch an average of roughly two hours a day, generating $10.5B in a single quarter at a roughly 50x P/E.

What replaces the add number is a three-line story: revenue, profit, and free cash flow. The leaked internal document suggests the trajectory the company is selling to itself. $78B in revenue by 2029 is roughly double the present run-rate. A $9B-per-year advertising business by the same year, layered on top of subscription price increases and the ad-tier rollout, does most of the work; the 410M subscriber figure circulating in the leak implies 36% growth from current levels. The market-cap target, $1T, is the punchline — the company would have to compound at a rate that requires advertising to scale, password-sharing crackdowns to continue converting, and the underlying engagement curve to keep bending.

Whether the leak was an accident is a fair question. Leaks of internal five-year plans to favoured outlets tend to be intentional when the planning is already public enough to make deniability cheap. The plan in question matches closely what analysts had been modelling after the ad-tier launch, the Microsoft 365 ad-tech partnership, and the live-sports push. The leak is not the news; the alignment of the disclosure change, the earnings beat, and the strategic narrative is the news.

The wider current: from growth to extraction

Netflix is the cleanest case study of a broader pattern that ran through the 16 June tape. Mature platform businesses are quietly migrating from the acquisition economy to the monetisation economy, and the reporting changes are arriving in lockstep with the strategic pivots.

Take the chip industry's parallel story, also surfaced this week. AMD, by the account of Dylan Patel of SemiAnalysis, has shifted into what Patel calls "wartime mode" to close Nvidia's CUDA software moat. The shift followed SemiAnalysis's December 2024 critique of the AMD software stack, and the changes are concrete: a developer-relations hire to run outreach, MI300X boxes shipped to well-known developers including George Hotz, and a public posture change from "never admit flaws" to a humility pivot that Patel argues is necessary if AMD is to replicate what Jensen Huang has done for a decade — donate high-end hardware to academic labs at CMU, Berkeley, UCSD, and seed the next generation of CUDA-native researchers. The 30%/70% cost structure that Nvidia's software moat is built on is being attacked at the developer-evangelism layer, where it is most expensive to defend.

Or take the defence budget. Delian Asparouhov of Founders Fund previewed the Hill and Valley Forum in Washington, pitching the gathering as the unlikely nexus where AI capital meets the appropriations process. His argument, made on the same broadcast, is that the reconciliation bill could add $150B–$200B to defence on top of the existing base, putting the next fiscal year near the $1T mark — a figure that is no longer treated as fringe. The spending priorities are nuclear modernisation, the Golden Dome missile shield, hypersonic boost-glide, and shipbuilding. The audience in the room, by the published attendee list, includes Jensen Huang, OpenAI's Kevin Wheel, Alphabet's Ruth Porat, and Interior Secretary Doug Burgum, with a space panel featuring SpaceX's Tim Hughes, Representative George Whitesides, and the DIU's Major General Bucky. The cocktail-hour diplomacy Asparouhov describes is the working theory: tech now needs permitting, energy, and reindustrialisation policy, and the cheapest place to buy it is the appropriations cycle.

The robotics story is the same story in a different ledger. Saman Farid, the CEO of Formic, argued on the broadcast that US reindustrialisation is bottlenecked by engineering capacity at small and mid-size factories, not by capital. His numbers are sharp. A typical US factory runs less than 2,000 hours a year out of a possible 8,700; a typical Chinese factory runs 7,500. China installed roughly 400,000 robots last year; the United States, despite much higher labour costs that should rationally justify more automation, installed about 30,000 across the entire installed base. Formic's wager is that 30% of a robot work cell is hardware and 70% is integration — and that the integration is the part software compresses. "Less than 5% of my deployment cost is programming the robot to do the thing," Farid said. "It's completely delusional how many people are spending their time trying to make the robot a little bit easier to program."

A different kind of capture, and a different kind of correction

The most provocative thread on the same broadcast belonged to Callie Means, a White House special government employee and co-founder of True Med, who framed the federal food-dye phase-out as "a correction of corruption" rather than nanny-state regulation. Her case: in the early 1990s, she argued, roughly half of the US food supply was controlled by Philip Morris and RJ Reynolds; the food pyramid was designed to push ultra-processed categories; American children now derive about 70% of their calories from ultra-processed food, against roughly 15% in Japan and Italy; the United States spends roughly five times what Spain does per capita on healthcare and lives about six years less. "Obesity is not an Ozempic deficiency," she said. The Medicare rule that made GLP-1 agonists the frontline defence, she added, predates the current administration and does not fund dietary alternatives.

The structural read here is uncomfortable. If a mature platform stops counting the metric that built it, the change is about pricing power, not progress. If a chip challenger admits its software stack is the problem and starts shipping boxes to influencers, the change is about closing a moat it cannot out-engineer. If a reindustrialisation push runs up against a 2,000-hour factory-utilisation floor, the change is about whether the country can run its own plants. And if a phase-out of food dyes is sold as anti-corruption rather than public-health, the change is about whether the political system can name the capture it is unwinding.

The disclosure changes — the metric retirements, the humility pivots, the cocktail-hour diplomacy, the inverting cost structures — are the early signal that the second act of the platform economy will look less like growth and more like the slow, contested redistribution of margin, attention, and industrial capacity.

The $1T Netflix market cap, the CUDA moat, the $1T defence budget, the 400,000 robots not yet installed: all of these are the same negotiation, conducted on different ledgers, between a maturing private sector and a public sector being asked to underwrite the next phase.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://www.youtube.com/watch?v=nQwbQpdijaM