Skip to content 
Revenue beat, and cloud grew 82% in its best quarter on record, yet Google also posted negative free cash flow for the first time since 2004, a trade-off that sits at the heart of this earnings report.
Alphabet’s Q2 earnings landed with a mixed reaction from the market. Shares slipped about 3.5% in after hours trading even though the company beat estimates on nearly every headline number, which tells you something interesting is going on beneath the surface.
Every quarter, Alphabet gives investors a small puzzle to solve. This time, the pieces were: a business growing faster than almost anyone expected, and a spending bill growing even faster than that.
For twenty years, Google has been the hyperscaler everyone assumed could fund its AI ambitions out of its own pocket, since its search business throws off enough cash that it never had to choose between growth and financial discipline. This quarter, that changed. Let’s unpack both sides of the puzzle.
The number everyone wanted to see
Google Cloud has spent the last two years playing catch up to Amazon Web Services and Microsoft Azure. This quarter, it stopped catching up and started sprinting.
Cloud revenue jumped 82% year on year to $24.8B, far ahead of the 64% growth pace the Street had been expecting heading into the quarter. The acceleration has now held for four straight quarters, not just one lucky one.
| 35% Q3’24 | 30% | 28% | 32% Q1’25 | 34% | 48% Q3’25 | 63% Q1’26 | 82% Q2’26 |
That growth isn’t a one quarter blip either. Alphabet’s cloud backlog, the value of signed contracts still to be delivered, grew to $514B from $460B in just three months.
That matters more than the headline growth number, because it tells you demand is still building, not just showing up in this quarter’s invoice.
The surge is coming from enterprises signing up for both AI infrastructure, the raw compute, and AI solutions, the finished tools built on top of it. Alphabet also began booking revenue from direct sales of its own TPU chips for the first time, its answer to Nvidia’s GPUs, though most of that money is still a 2027 story.
Search held up its end too, growing 17% to $63.3B, just shy of expectations, while YouTube snapped a losing streak with 13% growth.
| Segment | Revenue | YoY growth | vs estimate |
|---|---|---|---|
| Google Search & other | $63.3B | +17% | Slightly below |
| YouTube ads | Beat est. $10.81B | +13% | Beat, snapped a losing streak |
| Google Cloud | $24.8B | +82% | Beat by >$2B |
| Total advertising | $81.6B | — | Beat $81.1B est. |
| Other Bets (incl. Waymo) | $383M | +2.4% | Missed $401M est. |
The number that spooked the stock
The flip side of that growth is the bill attached to it. Alphabet spent $44.9B on capital expenditure in just three months, more than double what it spent in the same quarter last year, and roughly double what it spent every single quarter for the past two years running.
Instead of holding its full year spending plan steady, management raised it again, to a range of $195B to $205B, up from the $180B to $190B it had guided just last quarter. It’s the second upward revision to capex guidance this year alone.
| 13 Q3’24 | 14 | 17 | 23 Q1’25 | 24 | 28 Q3’25 | 36 Q1’26 | $45 Q2’26 |
The reason behind it is fairly simple: demand is outrunning supply. Google is still operating in a supply constrained environment, and it plans to lean more on third party cloud providers like CoreWeave and Nebius through Q3, using them as a bridge while it builds out its own data centres. Both those stocks jumped 4 to 5% on the news.
Alphabet is also renting Nvidia chips from SpaceX for roughly $920M a month just to keep up with demand.
| Metric | Q2’25 | Q3’25 | Q4’25 | Q1’26 | Q2’26 | Q2’26 YoY |
|---|---|---|---|---|---|---|
| Operating cash flow | 27,747 | 48,414 | 52,402 | 45,790 | 39,069 | +41% |
| Capex (property & equipment) | (22,446) | (23,953) | (27,851) | (35,674) | (44,924) | +100% |
| Free cash flow | $5,301 | $24,461 | $24,551 | $10,116 | ($5,855) | NM |
| Trailing 12-month FCF | $66,728 | $73,552 | $73,266 | $64,429 | $53,273 | -20% |
Looked at quarter by quarter, the shift becomes easier to see. Operating cash flow is actually still growing, up 41% year on year to $39.1B, which means the core business is throwing off more cash than ever.
The real problem is capex, which has more than doubled to $44.9B and has now outgrown operating cash flow altogether. That’s why trailing twelve month free cash flow, the steadier way to read this, has fallen 20% year on year to $53.3B even as revenue keeps climbing.
| Metric | Q2 FY26 | Q2 FY25 | Change |
|---|---|---|---|
| Revenue | $120.0B | $96.4B | +25% |
| Operating income | $40.8B | $31.4B | +30% |
| Operating margin | 34% | ~33% | +~100 bps |
| Net income | $112B | ~$28B | Quadrupled |
| Capex | $44.9B | $22.4B | +100% |
| Free cash flow | -$5.9B | Positive | Turned negative |
It’s worth pausing on that net income number, because quadrupling doesn’t mean the underlying business got four times better overnight. A large chunk of that gain came from Alphabet’s stakes in companies like SpaceX rising in paper value this quarter.
Operating income, which strips out those investment gains, is the more honest read on the business, and it still grew a healthy 30% to $40.8B, with operating margin expanding to 34%.
What about the AI model race itself?
This is the quieter worry sitting underneath the numbers. Google delayed the launch of Gemini 3.5 Pro this year, even as Anthropic and OpenAI kept shipping upgrades and Chinese open weight models kept getting cheaper and better.
Coding and agentic tasks are an area Google has openly acknowledged it still needs to improve, even as it maintains that it remains at the frontier on most other fronts.
The bigger story is what’s coming next. Google has already started training Gemini 4 and is pouring compute into it, treating it as a deliberately ambitious effort aimed at matching wherever the frontier will be by the time it ships, with future model releases expected to come at a faster pace.
On the consumer side at least, the Gemini app is already at 950 million monthly users, processing 22 billion tokens every minute. The underlying message from the company is that this still feels like the early innings of a much bigger shift across the economy, and that conviction in that opportunity has only grown over the past year.
The takeaway
Alphabet’s core businesses are all growing faster than expected, and its cloud backlog suggests that growth has real staying power rather than being a one quarter spike.
The trade-off is that Google no longer looks like the cash-rich, low-risk hyperscaler investors were used to. It’s borrowing, raising equity, and spending tens of billions a quarter to keep pace, just like every other AI-era tech giant.
The market isn’t questioning whether the AI bet is working. It clearly is. It’s questioning how long Alphabet can keep spending at this pace without its cash flow buckling under the weight of it, and for now, investors are choosing to wait and watch rather than celebrate.
So is this the AI bubble everyone keeps talking about? What makes Alphabet’s case different from a pure bubble narrative is that its incremental spending is chasing demand it can already see and price, that $514B backlog, rather than spending on faith that demand will eventually show up.
But that distinction doesn’t make the company immune to the industry-wide risk. Hyperscalers are now spending 45 to 57% of revenue on capex, a ratio that used to be unthinkable for technology companies, and a growing share of that commitment sits off balance sheet in signed but not yet started data centre leases.
If AI monetization growth slows even slightly while capex keeps compounding at these rates, free cash flow could stay negative for longer than one quarter, and that changes the calculus on the debt and equity Alphabet just raised to fund all this. The next two or three quarters of free cash flow, not this one, will likely tell us which story is actually true.
