Amazon has already invested a record amount in AI data centers, and the expenditure scale next year is expected to be even larger.
The world's largest cloud computing company announced its third-quarter results on Thursday and reaffirmed its capital expenditure plan for 2026 of $220 billion.
Amazon's spending has become a core issue looming over the company. To meet the surging demand for AI, the company is building large-scale data centers and purchasing expensive chips and other equipment.
This has pushed its free cash flow into the negative range. The third-quarter results show that there was a free cash flow deficit of $42.4 billion over the past 12 months.
Chief Financial Officer Brian Olsavsky stated that in the third quarter alone, Amazon's cash capital expenditures reached $75.9 billion, mainly invested in AWS and AI infrastructure.
This performance did not significantly alleviate investors' concerns about expenses. Amazon's revenue and profits in the third quarter fell short of Wall Street's expectations, and its stock price dropped by about 5% after the market closed.
eMarketer Chief Analyst Sky Canaves stated in a statement: "The improvement in the operating profit margin of cloud services is a welcome sign, but it may not be sufficient to alleviate market concerns regarding the sharp surge in capital expenditures and the corresponding dramatic decline in free cash flow."
Amazon has always argued that it takes several years for the AI infrastructure to build up and start generating revenue, whereas the AI chip servers can typically achieve break-even in less than three years once deployed. However, such large-scale investments also increase the market's expectations for the continuous rapid growth of its cloud business, especially when competitors are making similar substantial bets as well.
On Thursday, Jia Xi reiterated his confidence in the company's AI business model and prospects, stating, "We clearly see a path to achieve attractive financial returns."
The time it takes for Amazon to convert newly delivered servers into revenue-generating capacity is also under external scrutiny.
Business Insider reported exclusively last week that Amazon has begun receiving NVIDIA server racks worth billions of dollars, but due to power shortages and construction delays that limit new production capacity, not all of the equipment can be immediately deployed to the final data center sites.
"We are able to provide funding for this project."
Jia Xi said on Thursday that Amazon has a clear understanding of the demand that supports its large-scale AI investments. He mentioned that customers requiring the largest scale of AI deployment will reserve capacity well in advance, with capacities ranging from several hundred megawatts to even up to 1 gigawatt in some cases.
Jiaxi said, "Regarding the large capacities that customers hope to obtain, we have very good visibility into the demand for 2027 and 2028, and the same is true for part of the demand for 2029 as well."
In the third quarter, AWS's revenue increased by 39.3% year-on-year to $46 billion, marking the fastest growth rate in nearly five years. Jassy stated that Amazon expects AWS's revenue to grow by more than 40% in both 2027 and 2028, and this forecast requires the company to continue to rapidly increase its production capacity.
Jiaxi described this expenditure as a “one-in-a-lifetime opportunity” rather than an optional expansion. He said that it would certainly be more convenient for Amazon to spread this investment over 25 years, but the development speed of the AI market does not allow for such a approach.
He said, "We are able to provide funding for this project. It will permanently change our income and also permanently change our free cash flow."
The author of this article is Eugene Kim, and the source is Business Insider.
