Google’s search business just entered the compression phase

Alphabet reported Q2 2026 results this morning. Revenue up 24% year on year to $119.8 billion. Google Search & other at $63.3 billion. YouTube ads at $11.1 billion. Cloud growing at nearly a quarter per quarter. On any normal read, this is a monster quarter for a monster business.

And yet Alphabet posted negative free cash flow of $5.85 billion. They spent $44.9 billion on capital expenditure in a single quarter. They raised $49.6 billion through an equity offering to fund AI infrastructure buildout. Full-year capex guidance is $180–190 billion, six times what it was in 2022, and management has already signalled next year will be "significantly" higher.

Meanwhile — on the same day — the Wall Street Journal reported that USA Today, Politico, Reuters, People, The Economist, and possibly Reddit are all actively discussing whether to block Google's crawler entirely. Not throttle. Not paywall. Block.

Three stories. One structural situation. And the industry is largely treating them as unrelated news items.

They aren't.

The margin squeeze nobody is naming

Search & other grew 4.8% quarter-on-quarter. That's not bad. In a normal cycle, that's a healthy business. But the growth rate is decelerating from earlier quarters, and the composition of the growth matters more than the headline. Google is spending unprecedented amounts of capital to defend and extend the surface where that revenue is generated.

Google now needs Search to grow faster than it has ever grown, at exactly the moment its content supply is threatening to walk out the door.

The gap between operating cash flow ($39.1 billion) and capex ($44.9 billion) is the story. Alphabet is now, in cash terms, running the search business as a subsidised product of the AI infrastructure business. The old model was: Search prints money, we invest a fraction of it in the next thing. The current model is: Search prints money, we borrow against it and issue equity to fund the next thing, and the next thing costs more than the money-printer produces.

That is a different company. It's a company that needs Search revenue to keep growing hard enough to justify the capex profile it has committed to for at least the next 24 months.

Google now needs Search to grow faster than it has ever grown, at exactly the moment its content supply is threatening to walk out the door.

What the publisher revolt actually means

The Wall Street Journal piece is worth reading carefully because the quotes are more measured than the headline suggests. Nobody has blocked Google yet. Most of what's happening is publishers running the maths on what would happen if they did, and being willing to talk about it publicly for the first time.

That itself is the news. For twenty years, publisher relationships with Google were governed by an implicit deal: you let us crawl, we send you traffic. The deal held even when publishers hated it, because the alternative — invisibility — was worse than the compromise. That deal is now being renegotiated in public, and the renegotiation is happening because both sides of the trade have moved.

Traffic from Google to publishers has declined enough that "traffic is still traffic" — the Economist's Josh Muncke's phrase — is starting to sound like a resignation rather than a business case. And the value Google extracts from that content has increased, because that content is what makes AI Overviews and AI Mode possible. The exchange became lopsided. Publishers are now testing whether they can push back.

The interesting detail is Reddit. Reddit's content deal with Google — worth around $60 million a year — is up for renewal. Reddit has watched its Google visibility take real hits through the May core update and the June spam update. And Reddit is a company that now understands, more clearly than most publishers, that its content is training data as much as it is user-facing content. When Reddit's board considers whether to renew, they aren't just pricing referral traffic. They're pricing the value of their corpus to a company that needs it to compete with OpenAI.

If any single one of these publishers moves, others will move. Not because they've coordinated, but because the first mover creates permission for the rest.

The bind Google has walked into

Here is the structural position Alphabet is in as of Q2 2026:

Capex is committed. You cannot build a hyperscale AI infrastructure programme and then pause it because Q3 was soft. The data centres are being built. The GPUs are being ordered. The debt and equity have been raised. That number is going up next year regardless of what happens in search.

Search revenue needs to keep compounding. Cloud is growing faster in percentage terms, but Search is still the majority of the business by a wide margin and the highest-margin revenue Alphabet has. If Search decelerates, the capex programme starts to look reckless to public markets.

The AI features that Google is spending the capex on are the same features that are compressing referral traffic to publishers, which is the same reason publishers are considering blocking Google, which would reduce the corpus that trains and grounds those AI features.

That's the loop. And Google built it.

The optimistic Google read is that AI Overviews and AI Mode increase Search revenue over time by expanding query volume, by keeping users in Google's surface, and by opening new commercial formats. There's evidence for this — Search revenue is still growing, YouTube grew 12.2% quarter-on-quarter, and Google's investor commentary is confident. But the growth needed to justify a $180 billion capex year and a bigger one after that is not the growth of a mature ad business. It's the growth of a business that assumes AI transforms the product economics, not just the product surface.

That assumption depends on continued access to the open web's content — and on publisher willingness to keep supplying it under terms that are, from the publisher's side, getting worse.

Why this isn't cyclical

The reflex read on any single quarter is "cyclical." Ad markets go up and down. Capex intensity waxes and wanes. Publisher grumbling is a permanent feature of the internet.

What makes this different is that the three pressures are structurally linked.

The infrastructure spending isn't cyclical because it's a bet on a specific technology transition — one Google can't step back from without ceding ground to OpenAI, Anthropic, and Meta. The publisher revolt isn't cyclical because it's a response to a persistent shift in traffic dynamics, not a temporary dip. And the search revenue growth pattern isn't cyclical because the product itself is being restructured in ways that change per-query economics.

You can see all three moving in the same direction across the last four quarters. Capex up. Publisher patience down. Search growth decelerating slowly from a very high base. That's a compression, not a wobble.

Duane Forrester made an adjacent point this morning about attribution: OpenAI has shipped a full server-to-server conversions pipeline for advertisers and left organic practitioners with on/off toggles. Same signal, monetised on one side, starved on the other. It's the same 2011 "not provided" playbook, and it tells you something about where the platforms think the pressure has to be resolved. If Google is spending $180 billion a year on AI infrastructure, the monetisation gets built on the ads side first. The organic side gets whatever's left.

That's what a squeeze looks like from the inside.

The counterargument, taken seriously

The obvious pushback: Google has been here before. Every few years someone declares that publishers are about to revolt, that Search is about to be disrupted, that the capex programme is unsustainable. Every few years, Search grows anyway, publishers keep publishing, and Google's dominance compounds.

That read has been right for twenty years. It could easily be right again.

Where I'd concede ground: the publisher threat is easy to overstate. Even USA Today's chief executive, saying "enough is enough," hasn't actually blocked anyone. The Economist explicitly said blocking isn't currently on the table. Reuters is "weighing" it. This is publicly-visible negotiation posture as much as it is imminent action. Google has enormous leverage — traffic that's declining is still traffic that keeps some newsrooms alive — and every publisher blocking Google alone reaps roughly zero benefit unless enough others do the same.

And on capex: Alphabet has the balance sheet to sustain this for years. Negative free cash flow in a single quarter is not distress. It's a choice. The equity offering shows that public markets are still willing to fund the buildout at scale. Nothing about the numbers themselves is precarious.

Where I don't concede: the composition of the situation is genuinely new. It's not that any one of these pressures is unprecedented. It's that all three are showing up together, in a business that has, until this year, been able to solve each one in isolation. Google could always spend more capex because Search paid for it easily. Google could always tolerate publisher grumbling because publishers had nowhere else to go. Google could always afford Search deceleration because it had two decades of margin to burn.

The current configuration removes some of that slack from all three at once. That's what makes it a compression.

What this actually changes for the reader

If you're running SEO for a UK business, this doesn't mean Google is about to collapse. It doesn't mean you should stop caring about ranking. It doesn't mean the search channel is dying. Anyone selling you that story is selling you something.

What it does mean, practically:

The distribution channels are pluralising faster than most strategy decks acknowledge. If large publishers do start blocking Google — even a few of them — the shape of the SERP changes. AI Overviews get thinner in certain categories. Reddit and forum results become relatively more prominent in others. Direct-to-brand traffic patterns shift. The venues you optimise for are becoming less stable, not more.

Google's commercial pressure will show up in the product. Higher ad density in AI Mode. More sponsored placements. More Google-owned surfaces sitting between the query and the click. The self-citation pattern I wrote about earlier this month isn't an accident. It's what a compressed business looks like when it needs to preserve monetisation.

Brand becomes more, not less, important. When the platform layer is under pressure, the businesses that survive it are the ones that don't depend entirely on that platform for demand. Being known — being searched for by name, cited by name, recommended by name — is the moat that holds up when the pipes get renegotiated. This is the argument I keep making because I keep watching it be right.

And the measurement problem gets worse before it gets better. If publishers start blocking, if Google restructures AI Mode monetisation, if attribution windows narrow further — the instruments most SEO teams rely on will decouple further from reality. The measurement layer is already broken. This quarter's numbers just tell you which direction it's breaking in.

The bit worth saying plainly

Google spent more cash than it made this quarter, deliberately, because the AI infrastructure programme is now the priority above quarterly free cash flow. The publishers that supply the content that trains and grounds Google's AI products are, for the first time in the company's history, publicly discussing whether to walk away. And Search revenue growth — the thing that has to keep going up for any of this to make sense — is decelerating.

Any one of those things, on its own, is a story. All three of them, in the same 24 hours, in the same company, is a structural situation. The industry read of the earnings — "Google is fine, revenue is up, don't panic" — is technically true and analytically thin. Google is fine at the level of the current quarter. It is under compounding pressure at the level of the next five.

The businesses that will do best in this environment are the ones that stop planning as though Google is a stable platform and start planning as though Google is a company under strain. Because it is. The strain isn't fatal. It probably isn't even visible in the share price for another year. But it's real, and it will shape every product decision Google makes from here — including the ones that affect whether your content gets seen, cited, or clicked.

That's what the numbers said today. The story just isn't the one the headlines ran with.