Last night, after the US stock market opened, four major US companies—Microsoft, Google, Meta, and Amazon—simultaneously released their latest quarterly earnings reports. This was the first quarter in which the "Mag 7" companies released their reports almost on the same day, and the market was initially prepared to receive four reports that were "all good." Revenue and EPS did indeed all exceed analysts' consensus expectations. However, the market reaction cleanly divided them into two groups: Google surged 7.24%, Amazon rose slightly by about 1.3%, while Microsoft and Meta fell by about 2.4% and 6.6%, respectively.
These four companies have a combined market capitalization of approximately $12 trillion, accounting for more than a quarter of the S&P 500's weighting. When four earnings reports were released on the same night, the market should have been forced to make a judgment on a common variable. Instead, the result was not a unified direction, but a clear differentiation based on "conversion progress" in pairs.
"Exceeding expectations" is no longer valuable.
The revenue growth rates for all four companies were between 1.5% and 2.7%. Google's revenue was $109.9 billion, exceeding consensus expectations by 2.71%. Amazon's was $181.5 billion, exceeding expectations by 2.37%. Microsoft's was $82.9 billion, exceeding expectations by 1.84%. Meta's was $56.3 billion, exceeding expectations by 1.48%. These are not four "barely meeting" earnings reports, but rather four consistently strong earnings reports.
EPS figures appear even more impressive. Google's GAAP EPS was $5.11, exceeding expectations by 91%. Amazon's EPS was $2.78, exceeding expectations by 70%. Meta's GAAP EPS was $10.44. Microsoft's adjusted EPS was $4.27, exceeding expectations by approximately 5%. However, EPS cannot be directly compared horizontally. Google's figures contain a distortion of $36.9 billion in unrealized equity gains, Meta's include an $8 billion one-time income tax credit, and Amazon's high beat is mainly due to analysts' conservative estimates of AWS profit margin improvements. Therefore, this chart uses revenue beat to consistently assess the extent of "exceeding expectations" to ensure all four companies are on the same scale.
The strange thing is this benchmark. Amazon's beat was the second largest, but it only saw a modest 1.3% gain in after-hours trading. Google's beat was the third largest, but its after-hours gain was the largest at 7.24%. Meta's beat was the smallest, and it also fell the most. There's no simple logic that "the bigger the beat, the bigger the rise."
In other words, the market wasn't trading in the quarterly results that night, but rather in the two tables that followed the quarterly results.
$710 billion, a record Capex
According to the 2026 capital expenditure guidance, all four companies are increasing their spending.
Microsoft locked its 2026 calendar year capex at $190 billion during its earnings call. CFO Amy Hood explained that approximately $25 billion of that is incremental costs due to high memory prices. Previously, the consensus estimate for the Visible Alpha was only $154.6 billion. Overnight, Microsoft added a $35.4 billion spending forecast to the market.
Meta raised its full-year capex range from $115-135 billion to $125-145 billion, an overall upward shift of $10 billion. The CFO attributed the upward revision to "rising component prices" and "preparing for future capacity increases." The same financial report also disclosed a sequential decline in daily active users (DAU). These two factors combined create a situation of "spending more aggressively now, but growth momentum is weakening."
Google revised its forecast range from $175-185 billion to $180-190 billion, an overall upward shift of $5 billion. This is the most restrained of the four companies that revised their forecasts. The CFO also hinted that capex will continue to "significantly increase" in 2027.
Amazon maintained its $200 billion guidance from February. However, Q1 actual capex reached $44.2 billion, a 77% year-over-year increase. At this rate, it's highly likely to exceed the upper limit of its full-year guidance. Meanwhile, TTM free cash flow fell 95% from $25.9 billion last year to $1.2 billion.
Adding up the midpoints of the four companies' 2026 capex guidance amounts to $710 billion. $710 billion is exceptional even in the history of these four companies themselves.
In 2022, these four companies' combined capex was approximately $150 billion. In 2023, it remained largely unchanged, only jumping to $215 billion in 2024. From 2022 to 2024, the four companies collectively spent an additional $65 billion. 2024 was the real turning point. From that year onward, each year the four companies added another layer to their previous year's spending. The combined estimated spending in 2025 was $355 billion, almost double that of 2024. Then came the $710 billion in 2026.
This final leap was more dramatic than any previous year, with an annual increase of $355 billion from 2025 to 2026, equivalent to creating another expenditure volume equal to the entire year of 2025. In four years, Capex went from "each company building a few data centers a year" to "four companies combined building an entire country in a year".
This chart is reshaping the supply and demand of the entire industry chain. When Microsoft raised its capex, it specifically noted that $25 billion came from the "impact of high memory prices," and Meta also mentioned component price increases. The capex figure itself is growing larger and larger, but a significant portion of it is actually being drawn away from the upstream computing power chain: HBM, CoWoS packaging, electricity, land, and transformers—every single one is increasing. Spending $10 billion in 2026 will buy less computing power than in 2024.
What did the two companies that saw their prices rise do right?
Four companies are spending the same amount of capex, but only Google and Amazon have provided evidence this quarter that "money has begun to convert."
Google Cloud's revenue reached $20 billion this quarter, a 63% year-over-year increase, marking the first time Google Cloud has approached the scale of AWS and Azure in terms of size. Operating profit jumped from $2.2 billion in the same period last year to $6.6 billion, more than tripling. The backlog of unexecuted contracts nearly doubled quarter-over-quarter to $460 billion. This means that revenue visibility for the next few years has been significantly boosted.
During the same period, Gemini Enterprise's paid MAU increased by 40% quarter-on-quarter, and the total number of paid subscriptions for the entire company reached 350 million. The biggest bearish argument of the year, "AI is taking away search traffic," has been temporarily put to rest by this financial report.
Amazon, meanwhile, propelled AWS to a 15-quarter high. AWS revenue reached $37.6 billion this quarter, a 28% year-over-year increase, exceeding market expectations by 26%. AWS operating profit was $14.2 billion, exceeding StreetAccount's consensus estimate by 10%. Advertising revenue reached $17.2 billion during the same period, a 24% year-over-year increase, also exceeding market expectations. The simultaneous acceleration of these two high-margin businesses is key to the market's willingness to temporarily tolerate Amazon's $200 billion capex pace.
Comparing Microsoft and Meta, Microsoft Azure's growth rate of 39%-40% (at constant exchange rates) looks good on paper, but CFO Amy Hood directly told investors that Azure's computing power supply shortage will continue at least until the end of fiscal year 2026, that is, after June 2026. Customer demand has consistently outpaced supply, with the bottleneck being the speed of GPU and data center construction. Microsoft will have to wait several more quarters before converting capex into billable Azure revenue.
Meta's performance itself is impressive, but the overall upward revision of the capex range coupled with a sequential decline in DAU presents the market with a combination of "spending more money now, but front-end user engagement is actually loosening." This is the worst-received picture among the four companies.
The simultaneous release of earnings reports by four companies on the same day makes one thing clear: the ability to beat consensus expectations is no longer sufficient, and the market has begun to group the four companies according to their "conversion progress." Those who can convert capex into revenue or profit within the same quarter are rewarded. Those who can only present larger expense statements are penalized.
