Alphabet Earnings: Steady as She Goes.

In April we ran a series on the big-tech hyperscalers and argued that GOOGL 0.00%↑ was making the most balanced bet among peers relative to its overall business. At the time, they were paying for the entire AI build out of operating cash flow, while MSFT 0.00%↑ and META 0.00%↑ were already raising debt, cutting staff, and stretching depreciation schedules.

Given that the AI build-out is scheduled to run through at-least 2028, the question facing any hyperscalar is not whether they can cobble together gobs of cash now to get started, but instead whether they can sustain that spending for years to come. Ultimately, the levers being pulled by MSFT 0.00%↑ and META 0.00%↑ are available to Alphabet as well, and so one of the major questions going forward would be when, if ever, GOOGL 0.00%↑ pulls them. This question sits alongside how the core business holds up, as well as how the market revises its expectations of eventual AI payoffs.

So, let’s see how they did.

Did the self-funding hold? It did not. Though, arguably, Alphabet made no real attempt to make it. Capex doubled in the quarter, from $22.4B a year ago to $44.9B, and reached $80.6B across the first half. Operating cash flow didn’t grow fast enough to cover that, so free cash flow fell to $10.1B in Q1 and then to negative $5.9B in Q2.

To fill the gap, Alphabet went to the capital markets. Last month it announced what the WSJ pegged as roughly an $85B equity program. It actually issued $49.6B of stock and mandatory convertible preferred in June, with a $40B at-the-market facility behind it that it hasn’t yet tapped, and it added $20.3B of senior notes, all of which pushed long-term debt from $46.5B at year-end to $98.2 B.

When the dust cleared, the top line actually beat estimates, coming in at $119.8B against a consensus near $116.8B, but free cash flow turned negative. Perhaps signs of drifting closer towards Microsoft and Meta’s financial position? Or, to our argument in April, perhaps an appropriate increase in the investment’s aggressiveness.

Cloud is already turning AI investment into profit. Cloud revenue growth accelerated from +63% in Q1 to +82%, while segment operating margin went from 20.7% a year ago to 35.6%, and segment operating income more than tripled to $8.8 B. This is the scenario we sketched in April, when we wrote that if the supply constraints on compute eased, Alphabet’s investment could climb sharply and the profit engine would climb alongside it. Importantly, the demand underneath it is already contracted: Cloud backlog had already jumped to $467.6B at the end of Q1 from $242.8B at year-end. If a question about the health of Alphabet’s AI spending -- albeit a nitpick -- was whether it would convert into revenue growth quickly, cloud provides some evidence of a ‘yes’.

The data still says AI is helping search. The other half of what made the quarter reassuring is that the core business is as strong as ever. The nagging worry that AI Overviews would quietly hollow out Search is not showing up in the numbers. Search grew 17% to $63.3B, extending the re-acceleration that had carried paid clicks back to +13% in Q1 after five straight years of deceleration; management again pointed to AI Mode adding queries rather than eating them, with the Gemini app now at 950 million users. AI Overviews is past 2.5 billion monthly users. So far the click and pricing data say the transition is going smoothly, but this is an important second-order KPI to keep an eye on.

Anthropic is helping EPS gains, but is that a landmine? The reported EPS of $9.11, up 294%, is not what it looks like: $6.26 of it comes from a non-cash $99B gain on equity securities, almost certainly the Anthropic stake marked up again and why non-marketable securities nearly doubled to $131.5 B. Anthropic may be about to come under immense pressure from open-weight models, and should their valuation falter, it’s something that could dent Alphabet’s financial picture while the AI build-out is still underway.

R&D is up, and unlike peers, so is staffing. Shared R&D spending is up to a $5.8B loss from $3.4B, an annual run rate near $23 B. Staffing is also up, notably moving in the opposite direction from peers, with roughly 11,800 new employees to reach 198,933 over the same stretch in which Meta cut 8,000 to help fund its own build. Wiz, a ~$32B acquisition closed, appearing inside $33.7B of year-to-date deal spending.

While the overall financial picture looks strong, we expect some chatter about Gemini. Gemini 3 launched competitive with the frontier last November and has since slipped off it, the flagship 3.5 Pro release was delayed, and the company lost two key researchers to Anthropic in June. But with open weight models advancing at a rapid clip, the picture is much more complex that benchmark accuracy numbers. It is very possible that Gemini may become less important over time, and that Google is poised to profit from open-weight models within the cloud business. Our belief is that this is a point of interest, but not a meaningful weakness.

So, good Q2. Good work folks.

All Alphabet data sourced via #deepKPI from the company’s 10-K, prior 10-Qs, and the Q2 2026 earnings release, with the historical KPI series carrying one-click provenance to the underlying filings.