Amazon is launching the largest capital spending program in its history, trying to gain momentum against AI rivals by expanding data centers, developing chips and building models.
According to more than a dozen current and former senior employees, the group is undergoing a strategic makeover over fears that its cloud arm, AWS, is lagging behind competitors in securing corporate AI contracts.
Chief Executive Andy Jassy announced last week that Amazon’s capital spending will rise to $200 billion this year, more than that of Google and Microsoft, whose outlays will focus on computing infrastructure.
In December, he consolidated the group’s chip, model and advanced research teams under a single leadership structure, a move aimed at aligning its AI plans. The Amazon chief has also cut costs, including jobs — eliminating about 30,000 of its roughly 350,000 corporate roles.
“We have deep experience in understanding demand signals in the AWS business and then turning that potential into strong returns on invested capital,” Jassy said earlier this month. “We believe that will be the case here.”
AWS employees said the company’s moves also reflect internal concerns that it has failed to fully capitalize on its lead in cloud computing, particularly after OpenAI is slower than rivals to win major contracts with AI providers after it launches ChatGPAT in 2022.
“We were completely unprepared for how quickly things would unfold,” said a former senior AWS employee.
AWS remains the world’s largest cloud provider, generating nearly $130 billion in sales last year and more than 60 percent of Amazon’s total profits. But analysts predict that demand for AI-powered cloud services will cause Microsoft’s cloud business to overtake AWS in the next three years.
Amazon said other providers did not report “true cloud figures” and this prevented accurate comparisons between different cloud businesses.
About three-quarters of Amazon’s planned $200bn capital expenditure is allocated to AWS, according to public filings. Microsoft, Google and Oracle are on track to collectively spend nearly $400bn this year.
Jassy said last week that Amazon plans to add a meaningful amount of data center capacity this year. In 2025, it plans to add about 4 gigawatts of capacity – equivalent to the annual energy consumption of more than 3.2 million US homes. The group plans to double the capacity by 2027.
Investors are uneasy about the scale of Amazon’s bet. shares They are down more than 20 percent from their November peak amid concerns over how quickly spending will translate into returns.
Employees said pressure on Amazon to secure more deals with leading AI groups was among the main drivers of new cloud business. It has invested $8 billion in Anthropic and is building huge data centers for the group. But its initial equity investment came after Google backed the start-up.
Meanwhile, Microsoft, as one of OpenAI’s early investors, secured an exclusive cloud computing contract with the ChatGPT maker.
Amazon signed a $38bn cloud computing deal with the start-up last year only after Microsoft allowed OpenAI to undergo corporate restructuring. The deal dwarfs OpenAI’s $250bn contract with Microsoft and its $300bn worth of deals with Oracle.
Amazon said Friday: “It is not appropriate to infer that AWS was unable to secure major compute deals or was at a disadvantage when it comes to capacity planning. AWS continues to earn the majority of large enterprise and government changes in the cloud.”
It has announced the advancement of the group’s Graviton and Trainium chips, which are used for traditional cloud computing and AI training, respectively. Sales of these chips are expected to generate combined annual revenues of more than $10 billion.
Amazon introduced its latest generation of Trenium chips in December, promising significant increases in performance. People familiar with the matter said it is in talks to join OpenAI’s latest multibillion-dollar funding round, which is partly designed to ensure that ChatGPT makers adopt its semiconductors.
The chips should also help Amazon reduce its dependence on Nvidia’s products, which will help expand AWS’s profit margins from renting data-center capacity to corporate customers.
However, Google has attracted interest in its “Tensor Processing Unit” (TPU) by using its custom chips to power its Gemini AI model. Search giant sells 1 million TPUs to Anthropic in deal worth tens of billions of dollars.
Ben Bajarin of tech consultancy Creative Strategies questioned whether leading AI start-ups would adopt Amazon’s chips for their core products, even if they proved cheaper than Nvidia’s.
“Amazon specifically talks about price performance but the problem is that some users need absolute performance,” he said.
Amazon is also spending to advance its own “Nova” AI models, marketing them as a lower-cost alternative to rival models.
According to independent benchmarks, Nova underperforms the most advanced models created by OpenAI, Google, Meta, and Anthropic. According to three people familiar with the matter, executives have been irked by some AWS employees describing Nova as “Amazon Basics,” a term used for the group’s general home products.
Staff said the company is emphasizing its own AI tools like developer platform Kiro, while setting a goal for 80 percent of developers to use AI for coding tasks at least once a week. But many company engineers said they prefer to use Anthropic’s cloud rather than Nova for coding work, with one AWS engineer saying: “I didn’t even know we had a model.”
Amazon said its Nova model was used by “thousands of AWS customers” and performed comparable to some of the flagship models. It says AI labs like Anthropic are also deploying Trenium chips in large numbers.
JC said last year: “We will continue to insist on operating like the world’s biggest start-ups – customer-focused, inventive, fast-moving, lean, and full of missionaries trying to build something better for customers.”
The pressure to regain place in the AI race is taking a toll on employees. Some said they feared Amazon could be approaching “day two” — a term used by founder Jeff Bezos in 2018 to describe a business in “stasis,” followed by an “excruciating, painful decline.”
“The culture has changed, but so has the world around us,” said a senior AWS engineer at the company’s strategic pivot. “We have to prove our worth.”
