#美股财报季 AI Investment Trends Seen in Tech Giants’ Earnings
Tech giants’ earnings reports have shown no signs of weakening AI demand, yet have repeatedly rattled overseas markets.
After the market closed on October 21 U.S. time, Meta released its earnings report. Its share price fell about 7% the next day, while semiconductor companies including AMD and SanDisk dropped more than 8%. The Nasdaq fell 2.60%, and the Philadelphia Semiconductor Index plunged 6.72%. On October 29 U.S. time, Alphabet plunged 12.64%, wiping out about $500 billion in market value in a single day. After the market closed that day, Amazon and Intel released their earnings reports, putting further pressure on their share prices.
On one side, cloud business revenue continues to grow and AI capital spending keeps increasing; on the other, tech stocks are undergoing repeated corrections.
A review of earnings reports and calls from several tech giants shows that demand for computing power remains robust, while supply bottlenecks, rising costs, and financial pressures are gradually emerging. Combined with inflation and liquidity pressures, investors are growing more cautious, and their expectations for tech earnings have become more “demanding.”
Computing Demand Keeps Growing; Tech Giants’ Investment Remains Elevated
This earnings season shows that tech giants have not slowed AI investment despite market turbulence. Meta raised the lower end of its 2026 capital expenditure guidance by $5 billion to $135 billion. Alphabet maintained its guidance at $195 billion to $205 billion, Amazon expects full-year cash capital expenditures of about $220 billion, and Microsoft expects capital expenditures to exceed $50 billion next quarter. Several companies also said they would continue investing heavily in AI infrastructure in 2027.
The confidence behind these huge investments comes first from continued growth in cloud businesses. In the third quarter, Google Cloud revenue grew 103% year over year, Microsoft Azure grew 48%, and Amazon AWS grew 39.3%, with all three accelerating from the previous quarter.
AI demand is also extending further into inference and agent applications. The reporter noted that increased computing consumption on the application side is driving cloud providers to keep expanding infrastructure. Revenue from Microsoft’s Foundry platform reached more than three times the level of the same period last year, while token usage on Amazon’s Bedrock platform approached 17 times the level of a year earlier.
For cloud providers, customer orders are even booked years in advance. Microsoft’s commercial remaining performance obligations reached $677 billion, while Google Cloud’s backlog reached $561 billion. Amazon CEO Jassy revealed that most of AWS’s new compute capacity for this year has already been reserved, and customers have also taken most of the new capacity expected in 2027.
However, rising orders also mean new construction pressures. UBS Global Wealth Management’s Chief Investment Office said this earnings season shows that cloud business revenue growth is still accelerating and AI demand has not weakened significantly, while insufficient computing supply has instead become a major constraint on growth for some companies.
“Cloud providers need to procure chips, build data centers, and install supporting power and network equipment in advance to gradually convert customer contracts into revenue,” a person working in the cloud business at a leading domestic provider told the reporter. Continued growth in demand for computing power is extending competition beyond chip procurement to the overall capacity to deliver infrastructure.
Orders Growing Faster Than Deliveries; Supply Tightness Spreads Along the Value Chain
A set of figures in Microsoft’s latest earnings report reflects the real-world constraints facing computing infrastructure buildouts.
The company’s capital expenditures in its latest fiscal quarter were $45.2 billion, below market expectations of about $51.7 billion. Microsoft CFO Hood explained that this was mainly because some third-party data center capacity had not been delivered as planned, rather than any deliberate reduction in investment. The projects are expected to come online in stages in the second half of the current fiscal year, and capital expenditures are still expected to exceed $50 billion next quarter.
Bottlenecks in computing infrastructure are no longer limited to GPU supply. Supporting areas including power, land, data centers, network equipment, and storage are also affecting the commissioning of new capacity. A review shows that, to ease supply pressures, Microsoft is increasing revenue by optimizing GPU cluster scheduling and making better use of existing CPU resources, while Google has started purchasing more computing resources from third parties. Google Cloud is also selling complete systems built around its TPUs, boosting revenue but also tying up funds in hardware procurement and inventory.
Rich Privorotsky, head of European equity trading at Goldman Sachs, said in a recent interview that the market is watching whether data centers can secure enough power, whether equipment can come online on schedule, and whether backlogged orders can ultimately translate into actual revenue.
This supply-demand imbalance is also spreading upstream along the value chain. Intel’s data center and AI business revenue grew 78% year over year in the third quarter, but tight wafer capacity and supplies of advanced packaging substrates are constraining server chip deliveries. Intel CEO Lip-Bu Tan said the company is accelerating capacity ramp-ups to meet strong demand from server customers.
In storage, as demand for AI servers increases, some capacity is being redirected toward server products, putting pressure on costs for traditional devices. SanDisk’s data center revenue in its latest fiscal quarter grew 61% quarter over quarter, while consumer revenue fell 29%. Similarly, AI demand has boosted orders across the server supply chain, while traditional devices such as PCs are facing pressure from rising storage prices. Tan said that higher storage prices and tight supply have increased the cost of complete PCs, and some OEMs are adjusting inventory.
Markets Reassess Returns on Investment; Cash Flow and Traditional Businesses Face Tests
With robust demand and rising orders, why do concerns about tech giants’ earnings remain difficult to dispel? “Changes in these tech giants’ cash flows are one of the market’s most important considerations,” a private equity investor told the reporter.
This earnings season shows a marked divergence in free cash flow among the four tech giants. Microsoft’s free cash flow in its latest fiscal quarter was $18.6 billion. Although down 27% year over year, it was the only one of the four to generate positive free cash flow; Meta, Alphabet, and Amazon had net outflows of about $2.1 billion, $9.1 billion, and $33.5 billion, respectively. “This is also the key reason the giants’ share prices have diverged despite similarly significant growth in their cloud businesses,” the person said.
Expanding construction spending is also changing how tech giants arrange their financing. On October 29, Microsoft priced a $25 billion bond offering, while Meta had earlier raised about $24.9 billion in net proceeds through a bond sale.
J.P. Morgan Asset Management believes AI investment is still expected to grow rapidly, but capital expenditures at some large cloud providers have already exceeded their cash flow. Companies are increasingly turning to bond markets for financing, and higher interest rates are also raising funding costs.
Meanwhile, AI technology is beginning to affect tech giants’ established businesses. For example, Google search advertising revenue growth slowed from 19% in the first quarter and 17% in the second quarter to 12% in the third. Google executives acknowledged on the earnings call that AI Mode is increasing search queries, but its monetization efficiency remains lower than that of traditional search. A research team at Danske Bank also noted that Google’s earnings missed expectations mainly because of its core business, not AI-related revenue.
“Traditional businesses such as search, advertising, and retail remain an important source of funding for tech giants’ continued expansion of AI investment. AI can improve the operational efficiency of these businesses, but it could also change existing traffic channels and business models,” a securities firm analyst covering the computer industry told the reporter. Whether growth in new businesses can offset pressure on traditional businesses also affects companies’ long-term profitability and share price performance.
Tech giants’ earnings reports have shown no signs of weakening AI demand, yet have repeatedly rattled overseas markets.
After the market closed on October 21 U.S. time, Meta released its earnings report. Its share price fell about 7% the next day, while semiconductor companies including AMD and SanDisk dropped more than 8%. The Nasdaq fell 2.60%, and the Philadelphia Semiconductor Index plunged 6.72%. On October 29 U.S. time, Alphabet plunged 12.64%, wiping out about $500 billion in market value in a single day. After the market closed that day, Amazon and Intel released their earnings reports, putting further pressure on their share prices.
On one side, cloud business revenue continues to grow and AI capital spending keeps increasing; on the other, tech stocks are undergoing repeated corrections.
A review of earnings reports and calls from several tech giants shows that demand for computing power remains robust, while supply bottlenecks, rising costs, and financial pressures are gradually emerging. Combined with inflation and liquidity pressures, investors are growing more cautious, and their expectations for tech earnings have become more “demanding.”
Computing Demand Keeps Growing; Tech Giants’ Investment Remains Elevated
This earnings season shows that tech giants have not slowed AI investment despite market turbulence. Meta raised the lower end of its 2026 capital expenditure guidance by $5 billion to $135 billion. Alphabet maintained its guidance at $195 billion to $205 billion, Amazon expects full-year cash capital expenditures of about $220 billion, and Microsoft expects capital expenditures to exceed $50 billion next quarter. Several companies also said they would continue investing heavily in AI infrastructure in 2027.
The confidence behind these huge investments comes first from continued growth in cloud businesses. In the third quarter, Google Cloud revenue grew 103% year over year, Microsoft Azure grew 48%, and Amazon AWS grew 39.3%, with all three accelerating from the previous quarter.
AI demand is also extending further into inference and agent applications. The reporter noted that increased computing consumption on the application side is driving cloud providers to keep expanding infrastructure. Revenue from Microsoft’s Foundry platform reached more than three times the level of the same period last year, while token usage on Amazon’s Bedrock platform approached 17 times the level of a year earlier.
For cloud providers, customer orders are even booked years in advance. Microsoft’s commercial remaining performance obligations reached $677 billion, while Google Cloud’s backlog reached $561 billion. Amazon CEO Jassy revealed that most of AWS’s new compute capacity for this year has already been reserved, and customers have also taken most of the new capacity expected in 2027.
However, rising orders also mean new construction pressures. UBS Global Wealth Management’s Chief Investment Office said this earnings season shows that cloud business revenue growth is still accelerating and AI demand has not weakened significantly, while insufficient computing supply has instead become a major constraint on growth for some companies.
“Cloud providers need to procure chips, build data centers, and install supporting power and network equipment in advance to gradually convert customer contracts into revenue,” a person working in the cloud business at a leading domestic provider told the reporter. Continued growth in demand for computing power is extending competition beyond chip procurement to the overall capacity to deliver infrastructure.
Orders Growing Faster Than Deliveries; Supply Tightness Spreads Along the Value Chain
A set of figures in Microsoft’s latest earnings report reflects the real-world constraints facing computing infrastructure buildouts.
The company’s capital expenditures in its latest fiscal quarter were $45.2 billion, below market expectations of about $51.7 billion. Microsoft CFO Hood explained that this was mainly because some third-party data center capacity had not been delivered as planned, rather than any deliberate reduction in investment. The projects are expected to come online in stages in the second half of the current fiscal year, and capital expenditures are still expected to exceed $50 billion next quarter.
Bottlenecks in computing infrastructure are no longer limited to GPU supply. Supporting areas including power, land, data centers, network equipment, and storage are also affecting the commissioning of new capacity. A review shows that, to ease supply pressures, Microsoft is increasing revenue by optimizing GPU cluster scheduling and making better use of existing CPU resources, while Google has started purchasing more computing resources from third parties. Google Cloud is also selling complete systems built around its TPUs, boosting revenue but also tying up funds in hardware procurement and inventory.
Rich Privorotsky, head of European equity trading at Goldman Sachs, said in a recent interview that the market is watching whether data centers can secure enough power, whether equipment can come online on schedule, and whether backlogged orders can ultimately translate into actual revenue.
This supply-demand imbalance is also spreading upstream along the value chain. Intel’s data center and AI business revenue grew 78% year over year in the third quarter, but tight wafer capacity and supplies of advanced packaging substrates are constraining server chip deliveries. Intel CEO Lip-Bu Tan said the company is accelerating capacity ramp-ups to meet strong demand from server customers.
In storage, as demand for AI servers increases, some capacity is being redirected toward server products, putting pressure on costs for traditional devices. SanDisk’s data center revenue in its latest fiscal quarter grew 61% quarter over quarter, while consumer revenue fell 29%. Similarly, AI demand has boosted orders across the server supply chain, while traditional devices such as PCs are facing pressure from rising storage prices. Tan said that higher storage prices and tight supply have increased the cost of complete PCs, and some OEMs are adjusting inventory.
Markets Reassess Returns on Investment; Cash Flow and Traditional Businesses Face Tests
With robust demand and rising orders, why do concerns about tech giants’ earnings remain difficult to dispel? “Changes in these tech giants’ cash flows are one of the market’s most important considerations,” a private equity investor told the reporter.
This earnings season shows a marked divergence in free cash flow among the four tech giants. Microsoft’s free cash flow in its latest fiscal quarter was $18.6 billion. Although down 27% year over year, it was the only one of the four to generate positive free cash flow; Meta, Alphabet, and Amazon had net outflows of about $2.1 billion, $9.1 billion, and $33.5 billion, respectively. “This is also the key reason the giants’ share prices have diverged despite similarly significant growth in their cloud businesses,” the person said.
Expanding construction spending is also changing how tech giants arrange their financing. On October 29, Microsoft priced a $25 billion bond offering, while Meta had earlier raised about $24.9 billion in net proceeds through a bond sale.
J.P. Morgan Asset Management believes AI investment is still expected to grow rapidly, but capital expenditures at some large cloud providers have already exceeded their cash flow. Companies are increasingly turning to bond markets for financing, and higher interest rates are also raising funding costs.
Meanwhile, AI technology is beginning to affect tech giants’ established businesses. For example, Google search advertising revenue growth slowed from 19% in the first quarter and 17% in the second quarter to 12% in the third. Google executives acknowledged on the earnings call that AI Mode is increasing search queries, but its monetization efficiency remains lower than that of traditional search. A research team at Danske Bank also noted that Google’s earnings missed expectations mainly because of its core business, not AI-related revenue.
“Traditional businesses such as search, advertising, and retail remain an important source of funding for tech giants’ continued expansion of AI investment. AI can improve the operational efficiency of these businesses, but it could also change existing traffic channels and business models,” a securities firm analyst covering the computer industry told the reporter. Whether growth in new businesses can offset pressure on traditional businesses also affects companies’ long-term profitability and share price performance.
