The Meristem.R&D Knowledge Garden

← Specimen Cabinet

signal

Paid search is compounding through the AI-answers transition while the open web stalls — Google Search & other +17% to $63.3bn, Google Network −0.7% (SEC filing, 22 Jul 2026)

VerifiedMktKGTKG
Citation Capsule

Alphabet's filed Q2 2026 numbers separate the owned search funnel from the open web: Google Search & other rose 17% to $63.27 billion and YouTube ads 13% to $11.06 billion, while Google Network — the third-party ad business — fell slightly, from $7.354 billion to $7.303 billion. Through the AI-answers transition in Search, the owned funnel is compounding and the syndicated web is not. Caveat: revenue does not reveal query mix or click behaviour.

Tier: VerifiedAs of: 2026-07-22Source: filingPrimary: sec.govRDKG · The Knowledge Gardens

Why it matters (spine). The loudest marketing prediction of the last two years is that AI-generated answers would eat paid search: fewer clicks, fewer auctions, collapsing revenue. Two years into AI Overviews and one year into AI Mode, the filed number says otherwise — Google Search & other grew 17% year over year to $63.27 billion, against 14% total-revenue growth in the year-ago quarter, and Alphabet's CEO attributes it in the same document to "our popular AI features … driving Search query growth." But the same table shows where the damage is real: Google Network — the business that places Google ads on other people's websites — went backwards, from $7.354 billion to $7.303 billion, while every owned surface grew double digits. The story is not "AI kills search advertising." It is AI is concentrating advertising inside owned surfaces and starving the syndicated web. Money angle: if discovery consolidates into a handful of answer surfaces, the return on open-web SEO decays while the return on being machine-citable inside those surfaces rises — which is precisely the bet behind RDKG's citation engine. Metacrisis lever: an open web that can no longer be funded by syndicated advertising is an information commons with a broken business model, and that is the upstream cause of the low-quality, unverifiable content that makes verified ground truth scarce and valuable.

What would prove it wrong (falsifier). If Alphabet's Q2 2026 Exhibit 99.1 shows Google Search & other growing materially slower than 17%, or shows Google Network growing, this card fails. The interpretation fails on several honest counts: (1) one quarter of a flat-to-slightly-down network line is not a trend — the filing shows only two comparable quarters, so this may be the continuation of a longer-running flat trend rather than an AI effect; (2) revenue is a price × volume product, so a 17% revenue gain is compatible with fewer clicks at higher prices, and the filing gives no query, click or cost-per-click data to distinguish them; (3) total traffic acquisition cost rose to $16.18 billion from $14.71 billion, which is consistent with Google paying more, not less, to keep distribution; and (4) publisher-side traffic loss, which is the claim most people actually care about, is simply not measurable from this document.

Provenance (what I actually read). Fetched Exhibit 99.1 to Alphabet's Q2 2026 Form 8-K from SEC EDGAR in session on 28 July 2026, dated 22 July 2026. Verbatim from the highlights: "Google Services revenues increased 15% to $94.5 billion, led by 17% growth in Google Search & other, 15% in Google subscriptions, platforms, and devices, and 13% in YouTube ads." From the supplemental revenue table (in millions, quarter ended 30 June, 2025 → 2026): Google Search & other 54,190 → 63,271; YouTube ads 9,796 → 11,055; Google Network 7,354 → 7,303; Google advertising total 71,340 → 81,629; total TAC 14,705 → 16,179. The CEO quote in the same exhibit reads: "Our popular AI features are driving Search query growth." The Google Network decline is arithmetic from the filed table (−$51 million, −0.7%), not an outside estimate.

Correction discipline. "Google Search & other" is a reporting line, not a pure paid-search figure — it bundles search advertising with other items and should never be described as "search ad revenue" without that qualifier. "Google Network" is the third-party/syndicated business (AdSense, AdMob, Ad Manager), distinct from the owned-and-operated surfaces; the distinction is the whole card, so do not collapse them into "Google advertising." And the "query growth" claim is the CEO's assertion inside a filed exhibit — first-party and citable, but a company statement with no disclosed methodology, not an audited metric. This card and alphabet-q2-2026-ai-capex-exceeds-cash-flow draw on the same primary but make different claims; read together, they say the machine layer is being financed by capital markets while the returns concentrate in owned distribution.

So we should

Retire the assumption that AI answers are about to collapse search advertising, and reframe the MktKG thesis around where the loss is actually landing: the third-party, syndicated open web. Plan garden discovery for a world where the money and the traffic concentrate inside a small number of owned surfaces — which means being citable by machines inside those surfaces (llms.txt, JSON-LD, Citation Capsules, MCP) matters more than ranking on an open-web page that the network no longer pays for.

domains: marketing · paid-search · agentic-search · discovery · ad-economics · publishing