Amazon (AMZN) has spent years convincing Wall Street that Amazon Web Services is more than a profitable side business attached to an online retailer.
Artificial intelligence may expand that edge, according to KeyBanc.
The firm reiterated its Overweight rating and $350 price target on Amazon, forecasting Amazon Web Services revenue growth of 38% year over year, slightly ahead of the 37% consensus estimate. KeyBanc sees Amazon Bedrock and the company’s custom artificial intelligence chips as important drivers, according to Investing.com.
The call became more interesting on Oct. 19.
In order to enable developers to run code created for the well-known PyTorch artificial intelligence framework on Amazon’s Trainium chips with nearly no rewriting, the company released TorchNeuron. The software ecosystem that discourages developers from switching hardware is one of Nvidia’s (NVDA) main advantages. Amazon’s stock increased by roughly 5% following the announcement, according to Reuters.
For investors, this means that KeyBanc’s $350 call is about more than just the expansion of the cloud business.
Amazon is attempting to profit from AI in three ways: by renting out processing capacity, by selling model access via Bedrock, and by using AI to increase the value of its massive retail business.
KeyBanc sees AWS keeping its AI momentum
KeyBanc’s prediction has a solid foundation thanks to Amazon’s most recent performance.
AWS revenue surged 37% to $42.2 billion in the second quarter, its fastest growth rate in 18 quarters. AWS operating income climbed to $16.6 billion from $10.2 billion a year earlier, according to Amazon.
CEO Andy Jassy said the company’s artificial intelligence business had exceeded a $25 billion annual revenue run rate and was growing at a triple-digit percentage rate. Amazon’s chips business separately exceeded a $25 billion annual run rate and was also growing at triple-digit rates.
These numbers contribute to KeyBanc’s prediction that AWS growth would stay at around 38% rather than rapidly declining as the company grows.
Amazon is using both strategies to assault the market.
Bedrock offers multiple artificial intelligence model providers to corporate customers. Amazon reported that hundreds of thousands of customers use Bedrock and that second-quarter spending on the platform exceeded all previous quarters.
Trainium comes next.
For clients concerned about the cost of training and operating large AI models, Amazon’s bespoke AI processors are intended to offer an alternative to Nvidia graphics processing units.
Software has always been the problem.
Nvidia’s CUDA environment has been the focus of years of development. Even if the competitor chip is less expensive, shifting workloads to a new CPU may entail significant technical effort.
TorchNeuron goes straight after the obstacle.
Reuters reported that more than 30 early users are already testing the tool across robotics, voice AI and other workloads, with Amazon saying some customers have achieved meaningful cost reductions by shifting work to Trainium.
Amazon’s $350 case now rests on a bigger AI payoff
Amazon has another AI business hiding in plain sight
KeyBanc’s argument doesn’t end with AWS.
The firm expects Amazon management to emphasize its artificial intelligence assistants and the commercial potential of Alexa for Shopping, according to Investing.com.
More Amazon:
- Amazon is coming for your quick Walmart run
- Amazon’s Alexa shopping tool faces its first big October test
- Amazon pulls a best-selling pain reliever again, and customers need to check their medicine cabinets
Since cloud computing and chips get the majority of Wall Street’s AI attention, it is easy to undervalue that retail viewpoint.
Earlier this year, Amazon created Alexa for retail by merging Alexa+ with Rufus, its retail assistant. Consumers may view pricing history, compare items, receive suggestions, and automate transactions.
Amazon says active users of its AI shopping tools nearly doubled year over year during the second quarter, while interactions increased more than fivefold. More importantly for investors, U.S. customers using Alexa for Shopping spend more than 40% more per order on average than shoppers who do not use it.
This transforms AI from a characteristic of technology into something that investors can quantify much more easily.
Retail sales may increase if an AI assistant helps Amazon consumers find more items and create larger baskets. Amazon may also boost advertising income if companies pay to participate in such AI-driven talks.
Amazon Ads says more than 300 million customers used its earlier shopping assistant Rufus in 2025. The company has since combined Rufus with Alexa+ and introduced advertising formats specifically designed for conversational shopping.
Google and Microsoft may sell AI infrastructure. Nvidia can sell the processors that power it.
Amazon can offer the infrastructure and immediately present artificial intelligence to customers ready to make a purchase.
Wall Street is getting more bullish on Amazon’s AI math
KeyBanc is no longer alone at $350.
On Oct. 19, Morgan Stanley named Amazon its top pick, raised its price target to $350 from $335, and maintained an Overweight rating, according to Investing.com. The firm sees roughly 40% upside to its target.
When it comes to AWS expansion, Morgan Stanley has grown even more assertive than KeyBanc.
The firm now expects AWS revenue growth of 47% in 2027 and 42% in 2028, helped in part by greater availability of Nvidia graphics processors. It also sees Amazon generating roughly $16.50 in earnings per share in 2028.
Those estimates underline an important change in the Amazon investment story.
After consumers spent many quarters optimizing their cloud expenses, investors spent years wondering whether AWS’s growth would get back on track.
The question of how much more profit Amazon can make as the need for AI in AWS, chips, advertising, and commerce increases is gradually taking shape.
The solution will be expensive.
In contrast to a $18.2 billion inflow a year earlier, Amazon’s trailing 12-month free cash flow swung to an outflow of $7.6 billion in the second quarter. A $66.1 billion year-over-year increase in property and equipment spending, mostly for artificial intelligence infrastructure, was the main cause of Amazon’s decline.
Investors should be aware of this tension.
AWS is expanding quickly, but so is the amount needed to sustain that expansion.
Amazon is hardly alone. S&P Global warned this week that major hyperscalers are increasingly using complex financing structures to support AI infrastructure, while the industry’s capital requirements continue to climb, Reuters reported.
Amazon’s $350 case depends on turning AI spending into profit
KeyBanc still regards Amazon as one of the most alluring mega-cap companies.
Despite minor adjustments for foreign currency impacts, the company’s profits predictions are still 4% and 3% over consensus, respectively, and the shares are valued at around 23 times its 2027 earnings estimate and 18 times its 2028 earnings. India’s Investing.com
That valuation argument becomes much more compelling if AWS continues to grow at close to 40%.
If capital expenditures increase faster than the returns from such investments, it loses some of its appeal.
I find it noteworthy that Amazon now has many ways to benefit from its AI investment.
Through AWS, it may rent computer resources from Nvidia and Trainium. Bedrock may charge businesses for using it to access models. Using bespoke silicon may reduce the cost of its infrastructure. Additionally, it may convince customers to add additional things to their shopping carts by using AI assistants.
The recent TorchNeuron revelation strengthens one especially significant component of that equation.
When it comes to creating the fastest microprocessor in the world, Amazon doesn’t need to outperform Nvidia.
It must make Trainium so excellent and user-friendly that AWS clients have a reliable alternative.
This might reduce Amazon’s reliance on Nvidia while providing cloud users with an additional incentive to remain on AWS.
Therefore, KeyBanc’s $350 goal hinges on more than just excitement for AI.
Amazon must demonstrate that its massive expenditures are generating higher profitability, quicker cloud growth, and more valued customers.
So far, Wall Street increasingly thinks it can.
