Big Tech AI Earnings: Who’s Spending, Who’s Proving It Pays

The Q2 2026 earnings season has flipped the conversation around AI spending. Microsoft, Meta and Alphabet have all reported, and investors are no longer cheering raw investment totals. They want to see returns. The three companies together are spending at a pace that could exceed $700 billion industry-wide in 2026, according to current guidance figures. Morgan Stanley projects the number could cross $1 trillion in 2027. The results so far show one clear winner, one company in serious trouble with its investors, and one stuck in the middle.

What happened

Company	Key Q2 2026 fact
Microsoft	Azure grew 43%; total revenue $90B, up 18%; Copilot seats 30M
Meta	Free cash flow down 91% to $784M; 2026 capex guidance raised to up to $145B
Alphabet	First-ever negative free cash flow quarter, burning $5.9B; capex guidance $195-205B
Google Cloud	Revenue surged 82% to $24.8B; first TPU chip revenue recognised
Microsoft free cash flow	Down 23% year-over-year; quarterly capex up more than 70% to $41B
Morgan Stanley estimate	Industry AI spend could cross $1 trillion in 2027

For roughly two years, Wall Street rewarded AI ambition without demanding proof. That era appears to be ending. This quarter, all three companies saw their capital expenditures jump sharply, and investors responded very differently depending on whether those outlays produced visible revenue growth.

Microsoft: the clearest case that AI spending can work

Microsoft had the strongest story to tell. Azure cloud revenue grew 43% in the April-June quarter, beating analyst expectations, and the company forecast 45% growth for the following quarter. Total revenue reached $90 billion, up 18% year-over-year, with earnings also ahead of Wall Street estimates.

The Copilot figures were particularly notable. Paid Microsoft 365 Copilot seats climbed from 20 million to more than 30 million in a single quarter, well above analyst estimates. Microsoft’s commercial cloud backlog rose to $678 billion from $627 billion, and the company stated the entire sequential increase came from enterprise customers outside the leading US AI model developers. That suggests demand is widening, not concentrated in a handful of AI labs.

CEO Satya Nadella also flagged a strategic shift. Microsoft, which leaned heavily on OpenAI’s models early on, is now building its own models and custom chips, while letting customers pick from multiple AI providers based on cost and performance. Nadella said those efforts are already delivering efficiency gains of up to 40%.

Even with quarterly capex jumping more than 70% to $41 billion and free cash flow falling 23%, investors sent the stock up more than 8% in after-hours trading. The spend is large, but the revenue evidence is there. Our earlier coverage of Microsoft’s full Q4 2026 results has more detail on the Azure and Copilot trajectory.

Meta: big spending, thin proof

Meta’s quarter looked very different to investors. Free cash flow in Q2 fell 91% year-over-year, dropping from $8.55 billion to just $784 million. The company then raised the lower end of its 2026 capex guidance from $125 billion to $130 billion and said total spending could reach $145 billion this year. The market reacted by pushing the stock down 10% in extended trading, according to Reuters.

Meta has not yet presented the kind of concrete AI revenue metrics that Microsoft offered with its Copilot seat counts and Azure figures. That gap between spending and demonstrable return is what investors are penalising right now.

Alphabet: strong cloud, shaky balance sheet

Alphabet’s numbers had two very different stories inside them. Google Cloud revenue surged 82% to $24.8 billion, well ahead of expectations, driven by enterprise AI computing demand. Total company revenue reached $119.8 billion and advertising also beat estimates. Alphabet also recognised revenue from direct sales of its Tensor Processing Units (TPUs, the custom AI chips Google designs in-house) for the first time.

But the cash flow picture was alarming. Alphabet posted its first-ever negative free cash flow quarter, burning $5.9 billion. It also raised its 2026 capex guidance to between $195 billion and $205 billion, a $15 billion increase from its previous forecast, citing AI infrastructure demand outpacing available capacity. Investors remained cautious despite the cloud beat, unsettled by rising costs and reported delays to its flagship AI models.

The tension inside Alphabet’s results is real: the cloud business is growing fast, but the cost to sustain that growth is accelerating faster than revenue.

Why it matters

This earnings season marks a clear shift in how the market judges AI investment. The question is no longer “are you spending on AI?” It is “what are you getting for it, and when?” Companies that can show concrete metrics, seat counts, cloud growth rates, backlog figures, are being rewarded. Companies that cannot are being punished, even if their long-term thesis is sound.

Apple, which largely avoided the infrastructure spending race, is being watched as a possible beneficiary if the market continues to cool on pure-spend stories. Amazon has not yet reported, and its AWS cloud results will be the next significant data point.

For businesses evaluating their own AI tool investments, the same logic applies. Spending on AI integration without measuring output is the same mistake at a smaller scale. The companies winning this cycle are the ones tying investment to measurable results.

Our take

Microsoft’s results are the most instructive here. The 30 million Copilot seat figure is a real number tied to a real product with a real price. That is what separates it from the vague “AI demand is strong” language coming from the others. When a company can say “we sold X seats at Y price and the backlog grew by Z,” investors listen.

Meta’s problem is not that it is spending. It is that it has not yet produced the equivalent clarity. A 91% drop in free cash flow is a serious signal, and raising guidance into a gap between spending and returns is a hard sell.

For our clients thinking about AI tools and workflow automation: the lesson from this earnings season is that AI spend without measurement is just cost. Before you expand any AI investment, make sure you have the equivalent of a “seat count” for your own use case. Something concrete you can point to that justifies the line item.

What to do about it

1. Audit your current AI tool costs against documented time or revenue outcomes before adding more subscriptions.
2. Track one concrete metric per AI tool (hours saved, leads generated, support tickets deflected) so you can make the same case to your own stakeholders that Microsoft made to its investors.
3. Watch Amazon’s AWS earnings for the next signal on whether enterprise AI demand is as broad as Microsoft claims.
4. If you are evaluating Microsoft 365 Copilot, the seat growth figures suggest enterprise adoption is accelerating. That is a reasonable time to pilot it, with clear success criteria set in advance.

The takeaway: AI spending is only a strategy if you can name what it buys.

Source: Bing News · Meta AI