Amazon Q2 Earnings: AWS AI Demand Outpaces Capacity Through 2027, Jassy Says

‘The demand we already have for 2028 is striking,’ Amazon.com CEO Andy Jassy said.

Amazon Web Services saw voracious demand for artificial intelligence also fuel growth in its core cloud business in its latest fiscal quarter while the ongoing inflated prices for data center components including memory adds an extra $20 billion to the amount of capital the vendor expects to spend in 2026.

“Even at that amount, we will still not have enough capacity to meet all the demand we have in 2026,” Andy Jassy, CEO of AWS parent Amazon.com, said Thursday during the vendor’s latest quarterly earnings call. “I believe this dynamic will also be true in 2027, too. In fact, the demand we already have for 2028 is striking.”

Jassy covered the Seattle-based cloud giant’s second fiscal quarter earnings Thursday, spanning the three months ended June 30.

Amazon Raises 2026 Spending Forecast To $220 Billion

Duane Barnes, president of Raleigh, N.C.-based solution provider and Cox company RapidScale—No. 173 on CRN’s 2026 Solution Provider 500—told CRN in a recent interview that his company is investing in its AWS practice and has achieved around 20 badges with the vendor, including one for agentic AI expertise.

“We have got a long history of moving thousands of clients into the public cloud,” Barnes said. “We’ve got eight advisory and consulting practices around infrastructure, cyber security, AI, ML, FinOps, so on and so forth.”

Barnes’ public cloud business has seen increased demand for the solution provider’s financial operations (FinOps) expertise as well as cloud migrations from legacy vendors. RapidScale’s approach has taken on a new level of importance as customers navigate higher prices for data center components including memory, servers and storage.

The goal is figuring out “where should these apps live long term—probably some mixture of public cloud and private cloud and on prem,” he said. “And then you can really get more life out of your infrastructure if you put everything where it should be.”

Memory Crisis Impacts Amazon Spending

Thanks to inflated memory and data center component prices, Amazon raised its expected 2026 spending to $220 billion, Jassy said on the call.

Jassy explained on the call that he feels comfortable with the large amount of capital spending because data centers offer 30-plus years of monetization without additional startup capital, with servers and networking equipment breaking even in less than three years, and customers usually contract AI capacity for at least five years. Jassy foresees five to six generations of those server economics, with subsequent generations having better economics without additional data center investment.

Jassy said that Amazon’s emerging AI business is “a little bit ahead” of the same margin trajectory as the core cloud business, also making the CEO feel comfortable with the high level of spending.

Amazon’s cash capital expenditures in the second quarter was $53.1 billion, mostly for AWS and generative AI.

AWS Revenue Jumps 37 Percent As AI Fuels Cloud Growth

Amazon saw growth in its own AI products and services while overall growth in AI inferencing helped AWS grow net sales in the quarter 37 percent year on year, the fifth straight quarter of accelerated growth and the fastest growth in 18 quarters, according to the vendor.

Its annualized revenue run rate hit $169 billion and it added $4.6 billion in revenue quarter on quarter, 80 percent above the previous largest increase.

That ARR would place AWS 24th on the Fortune 500 list if it were a standalone company, Jassy said on the call. AI is helping push AWS to a $1 trillion annual revenue business.

AWS segment sales grew 37 percent year on year to $42.2 billion. Backlog for AWS is at $496 billion, up triple digits year on year.

Jassy credited the growth to AWS’ broad functionality across core cloud use cases and AI, its operational performance, security and users choosing AWS for bringing inference workloads to production.

The CEO said that the classic cloud computing market still has room to grow, with 85 percent of global IT spending remaining on premises. “That equation is going to flip in the next 10-to-20 years,” he said. “You see more enterprises that are moving and building plans to move to the cloud, and we’re winning the lion’s share of those.”

Amazon AI Business Surpasses $25 Billion Annual Run Rate

Amazon’s AI business surpassed a run rate of more than $25 billion, marking triple-digit percentage growth year on year.

The CEO pointed to growing demand in the vendor’s own AI products and services, with Amazon Bedrock—the fully managed generative AI application-building platform—now serving hundreds of thousands of customers. Bedrock added more customers in the last six months than in the first two years after launch.

Customers spent more in the second quarter than all prior quarters combined, Jassy said.

Although Amazon’s AI business has been successful so far without a frontier AI model, Jassy said the vendor will continue to build one would allow greater control over costs, speed and models’ priorities.

“We’re trying to drive costs down for customers, and having a player like ourselves always focused on trying to take the price, performance and cost out for customers all the time we think will help keep the model more cost effective for customers,” he said. “Within the next few years, you’re going to have at least half a dozen models that are comparably good to each other. They’ll all be in Bedrock and one of them will be ours.”

Jassy attributed some of Bedrock’s success so far to customer desire for model variety and model choice.

“There is not going to be one model to rule the world,” the CEO said.

Amazon’s Kiro coding agent tripled in usage quarter-over-quarter, Jassy said. He opened the possibility for more AI applications by Amazon in the future, saying that “we see a very substantial opportunity both for our customers and for AWS in building some of these agentic applications.”

“There are several others that we’re working on, and we think it’s going to be very helpful for customers and our business,” Jassy said.

Trainium, Graviton Power Amazon’s AI Infrastructure Strategy

Like how Amazon is investing in a frontier model akin to ones offered by OpenAI, Anthropic and other leading AI research labs to help drive down costs, the vendor hasn’t slowed down on making its own semiconductors as well.

The Amazon chips business posted results like those of the AI division, with chips surpassing a run rate of more than $25 billion, marking triple-digit percentage growth year on year.

AWS reported more momentum in its Trainium AI chips during the quarter, with multi-year, multi-gigawatt commitments from Anthropic and its rival OpenAI.

Amazon’s Graviton5 Arm-based cloud processor has reached a 98 percent share of AWS’ top 1,000 Elastic Compute Cloud (EC2) customers, according to the vendor. Revenue commitments grew nearly threefold quarter on quarter. Graviton5 is growing nearly twice as fast as Graviton4 did and delivers up to 25 percent better compute performance.

AI growth has led to core cloud business growth thanks to post-training, reinforcement learning and agentic tool use on cloud-delivered central processing units (CPUs) as opposed to AI accelerators. This also bodes well for Amazon’s own Graviton CPUs, he said.

AWS is still on track to double the power capacity it had in 2025 by the end of 2027, with no AI demand slowdown in sight, the CEO said.

“We’re still in the relative early stages of how much demand there’s going to be for AI,” he said. “It’s going to change every customer experience that we know. I think it will invent all sorts of new ones that we never imagined.”

Asked about whether Amazon would sell Trainium chips to customers separate from its cloud, Jassy said that “we’re actively having those conversations and exploring, and I expect there’s a real chance we’ll do that in the future.”

Amazon’s investment in first-party chips helps make the vendor “unusually well positioned for the AI inflection,” Jassy said. The CEO also reiterated Amazon’s deep partnership with Nvidia, a semiconductor manufacturer that Amazon could someday potentially challenge with its own chips portfolio.

Amazon Q3 Outlook

Amazon still has an eye on bleeding-edge markets outside of AI, with the completion of four additional launches for its Leo low Earth orbit satellite network during the quarter. That brings the total constellation to nearly 400 satellites in orbit with satellite internet service expected to start this year.

Amazon reported that its second quarter net income of $62.6 billion, more than triple the $18.2 billion reported in net income for the same period a year prior, includes income from its investments in Claude maker Anthropic. That income “primarily” contributed to non-operating pretax other income of $53.4 billion.

For the third fiscal quarter, Amazon expects to see net sales between $197 billion and $202 billion in its third fiscal quarter, up 9 percent to 12 percent year on year. Those figures include all of Amazon’s businesses including its e-commerce division.

Operating income should land between $22.5 billion and $26.5 billion, up about 30 percent to 52 percent year on year.

Amazon’s stock jumped 9 percent in after-hours trading, with the stock trading at about $260 a share.