Amazon’s stock price had already been dropping in after-hours trading on Thursday despite better-than-expected results when Morgan Stanley analyst Brian Nowak prefaced his questions on an earnings call with a disclaimer that made it clear this wasn’t going to be a “Congrats on the quarter, guys” type of analyst—CEO interaction.
“I have two [questions] for you on AWS; they’re a little tough but I’m going to throw them at you,” Nowak told Amazon CEO Andy Jassy. “There is a Wall Street finance person narrative right now that AWS is falling behind in Gen AI with concerns about share loss to peers. What is your rebuttal to that and talk to us about your and the team’s most important focal points just to ensure that AWS stays on the knife’s edge of innovation versus hyperscaler peers?”
Nowak also pressed Jassy on why it wouldn’t be fair to assume that AWS’ revenue growth shouldn’t accelerate in the back half of the year given all of AWS’ generative AI offerings and widespread demand from companies of all sizes to cash in on this transformational technology.
Jassy responded by stressing that this is the early stages of a technological transformation that will extend far into the future. While some of the top frontier model providers do use AWS in some capacity, non-AI AWS customers that are rushing to build generative and agentic AI services using AWS are “quite early, and many of them are just smaller in terms of usage relative to some of those top heavy applications I mentioned earlier.” That is bound to change.
So if you follow Jassy’s thinking, as more enterprises figure out what they want to build and how they want to build it, they’re going to start having different needs. For the largest model makers, like Open AI or Anthropic, Jassy foresees their costs shifting from a mix between training their models and the cost associated with “inference,” or the customer-facing part where the model spits out a prediction, answer or action, to mostly inference expenses. And Jassy maintains AWS is positioned well for this transition because of the low-cost AI chip line Trainium.
“It’s about 30% and 40% better price performance than the other GPU providers out there right now, and we’re already working on our third version,” he said.
For others, who want to use another company’s model to create their own generative AI applications, Jassy argued that Amazon Bedrock, which offers models from a wide selection of companies, has become a go-to destination and “is growing very substantially.”
Jassy continued on the this-is-just-the-first-inning thread, by noting that companies are just starting to think about deploying AI agents and that, with its recent agentic AI announcements, AWS will be well-positioned to capitalize.
The Amazon CEO, and former AWS chief, added that AWS cloud leadership position also provides some lock-in as AI “inference” becomes just another component of a company’s cloud services stack.
“[P]eople are going to actually want to run those [AI] applications close to where their other applications are running, where their data is,” Jassy said. “There’s just so many more applications and data running in AWS than anywhere else.”
As for Nowak’s question about the possibility of AWS’ growth rate accelerating in the back half of the year, Jassy wouldn’t directly answer it but stressed his optimism, in part stemming from more AWS starting to deploy more AI products at sale that should continue to ramp in coming quarters.
Earlier in the call, Jassy had defended AWS’ 18% revenue growth rate in light of Microsoft reporting 34% annual revenue growth for its Azure cloud unit and Alphabet recently reporting 32% quarterly growth for Google Cloud. Azure generates around 2/3 the revenue that AWS does, while Google Cloud registers less than half the annual revenue of Amazon’s cloud behemoth.
“You look at the business, it’s a $123 billion annual revenue run rate business and it’s still early,” he said. “How often do you have an opportunity that’s $123 billion in annual revenue run rate where you say it’s still early? It’s a very unusual opportunity that we’ve very bullish about.”
Ebay turns 30 this September and the company is showing some signs that it’s starting to turn back the clock.
The auction giant’s stock surged more than 19% on Thursday to an all-time high of $92 a share on the back of revenue growth and earnings that both surpassed analyst expectations, and an outlook that was far less gloomy than it could have been.
That’s the top-line good news.
The reality check, though, is that Ebay still accounts for less than 4% of the overall e-commerce market in the U.S., compared to around 40% for Amazon, 6% for Walmart, and pressure from a collection of discount or product-specific apps and marketplaces from Temu to Whatnot, a live-streaming commerce startup valued at $5 billion.
So what gives? Where is the eBay optimism coming from and is this current moment an aberration and blip or, as the internet analyst Mark Mahaney wrote in a research note on Thursday, “something of an inflection point”?
As Mahaney wrote, eBay’s gross volume and revenue growth rates, excluding the impact of foreign exchange rate changes, both rose 4% year-over-year in the past quarter. Those numbers sound meager because they are, but it’s the fastest that the commerce marketplace has grown in two years—which is…something.
The one-time e-commerce darling has been trying to lean into its historic strengths or “focus areas”—product categories where it can win—and that seems to be paying off. Those categories grew 10% in the quarter compared to about 1% for everything else.
“Collectibles was once again the largest contributor to growth as year-over-year growth in trading cards GMV accelerated for the 10th straight quarter on the back of continued momentum in both collectible card games and sports trading cards,” CEO Jamie Iannone told analysts on the earnings call.
Collectible categories have been booming across the board since the onset of the pandemic, as nostalgia coupled with new shopping experiences and sales tactics drive renewed interest in trading cards like Pokemon as well as sports cards. Startups like Whatnot, Fanatics, and Courtyard, which is selling $50 million in cards and comics a month, are playing a role. But eBay is still the biggest online player in the space and also capitalizing on the consumer interest.
To that end, the company is pushing aggressively into live-streaming commerce, with eBay Live, where Fanatics and Whatnot are finding success, and the early signs are encouraging, according to Iannone.
“We have already seen significant evidence that live commerce can deepen engagement among eBay enthusiasts and unlock even greater velocity in our strongest verticals, which validates our continued investment in this experience,” he told analysts.
(For what it’s worth, top live-streaming card seller Rene Nezhoda, of Storage Wars fame, had this to say about eBay Live when I interviewed him about his business on Whatnot, where he sells $15 million to $18 million a year in sports cards. “I wouldn’t be surprised if eBay Live might become the market leader because they just have such a big user database,” he told Fortune in June.)
Ebay’s Iannone said his company is also seeing both top-line and bottom-line benefits from integrating the Klarna Buy Now Pay Later payment method into the shopping site, by expanding its on-site advertising business, and by integrating generative and agentic AI experiences into the shopping marketplace for both buyers or sellers, though it’s too early to tell how much the latter will boost the business long term.
No, eBay isn’t immune to the U.S.-China tariff war, though it shouldn’t be crippled by it. No, it’s not a threat to giants like Amazon and Walmart that specialize in everyday, new goods. But riding hot, historically strong product categories, while keeping up with new industry tablestakes like Gen AI integrations for sellers and buyers, and popular payment types like Klarna, has the 30-year-old company’s future looking brighter than it has in a long while. And that’s not nothing.
I’m going to start with a caveat from the top: This is not a formal product review. That’s not my background nor expertise, and if that’s what you are looking for, you are likely to walk away at least a little bit disappointed.
What this is, is a first impression based on hands-on experience with the new Alexa from someone who was once a consistent user of Amazon’s original voice assistant. Back then, I relied on Alexa for the kind of straightforward things many of us did every day: playing music, checking the weather, requesting sports scores, setting timers, and answering the types of questions that grade-school kids would get a kick out of (“Alexa, who would win a battle between a lion and a snow leopard?”). But over the years, Alexa’s performance seemed to deteriorate– it had more trouble understanding basic requests and definitely could not hold a conversation like popular AI chatbots could. Eventually, my family’s interest—and patience—waned.
So I’ve been waiting for a new and improved Alexa for quite some time, and when I recently received an invitation offering “early access” to the beta version of Alexa+, I was eager to take it for a verbal spin.
It’s worth noting that Amazon first announced what would become Alexa+ back in September of 2023, but the launch has been repeatedly held up amid “structural dysfunction and technological challenges,” as Fortune reported last June, and later by issues related to how slow the assistant was to respond to commands or complete actions. In February, Amazon finally unveiled details of Alexa+ at a splashy launch event, but did not launch the service widely at the time; instead, it’s been rolling out Alexa+ little by little, in a phased approach (Amazon says that millions of people now have access to Alexa+). Prime members don’t pay anything for the Alexa upgrade, but non-members will pay around $20 a month after the official launch, the company has said. For now, early access is free to Prime and non-Prime members alike. The company has not formally announced an official full public launch date.
I’ve spent some time over the last few weeks using Alexa+ for some of the same things we used its predecessor for, as well as trying out some of the new actions, like booking an Uber and restaurant reservation, that Amazon is pushing. My first impression, in short, is that the service is pretty good. If it had launched shortly after Amazon first announced an updated version of Alexa in the fall of 2023, I might have said it was very good. Its conversational abilities are real and mostly very fluid. Does it blow away voice modes from LLM-based AI assistants like ChatGPT and Perplexity? Not in my experience. But it is vastly superior in that way to the original Alexa so will likely come as a delightful surprise to those who haven’t spent much time with those competitor services. On several occasions, though, I had to re-prompt Alexa by name in the middle of a back-and-forth conversation—I thought I had just taken a normal, mid-speaking pause but Alexa thought differently. If such instances continue to occur at public launch, it might not be a deal breaker for regular usage, but would certainly frustrate me – and I assume some others too.
Can you hear the music playing?
I also had some issues with playing Spotify using the new Alexa, unless I specified that I wanted it played on the specific Alexa device in front of me. The gadget in question was an 8-inch Echo Show device (the Echo device with a screen) to test out Alexa+ because the technology isn’t available on some of Amazon’s older speakers, including the original Pringles-box-shaped Echo speakers, one of which still sits on a shelf in our dining room. (If you don’t have an Echo device, you’ll still be able to use the new Alexa+ from the Alexa app.)
Earlier versions of the Echo smart speaker looked like cylindrical Pringles chip boxes. The new Alexa+ is not compatible with some of those earlier versions.
Cayce Clifford/Bloomberg via Getty Images
The new Alexa told me Spotify was playing, when it actually wasn’t. I thought perhaps it was somehow playing on the old Pringles-tube Echo downstairs, but that wasn’t the case. A spokesperson recommended I change the default device for Spotify in the Alexa app but honestly, the Alexa app isn’t the most intuitive and I gave up after about 10 minutes. Considering that playing music is one of the basic and common tasks for a smart speaker, this didn’t inspire a lot of confidence, but I am not ruling out the possibility that I’m overlooking a setting that would fix the issue.
The other flaws I ran into ranged from comical to frustrating. An on-screen prompt on the Echo Show advertised that Alexa could help me choose a new lunch spot, but when I queried Alexa about it the first time, she claimed she couldn’t carry out that task.
I also made the mistake, apparently, of asking Alexa to slow down her speaking cadence at some point so I could take some hand-written notes. That simple command kicked off a minutes-long bizarro-world exchange in which I would ask Alexa to speed up or slow down her cadence, she’d reply that she had—but at a speed which was even more drastically opposite of what I had been asking. It took several minutes, but what felt like an eternity, to rectify.
On another occasion, Alexa got snippy with me when I seemed astonished that she had instructed me to simply unplug and then reconnect my Echo device to try to solve the aforementioned Spotify issue. “It’s your problem not mine” was essentially the gist of the response. Can an AI offend me? I mean, that’d be pretty silly. But the exchange was a bit off-putting, though admittedly mildly amusing as well.
On this point, Panos Panay, the longtime Microsoft executive who joined Amazon in late 2023 to head up Alexa and its broad array of devices from Echos to Kindles to Fire TV sticks, seemed intrigued.
“We’re testing a few of the boundaries,” he told me in an interview at the company’s New York City headquarters in early July. “Like, yeah, you want a little personality out of your assistant, and you want it to feel or be personal. I think that’s okay. Where is that boundary is an interesting question.”
Alexa’s new tricks
For my daughter, Alexa+’s ability to generate images and “paintings” based on voice commands was a treat. I also tried some of the advertised “actions” that Panay and Amazon believe will set Alexa apart from competitors and transform it into more of an agent than an assistant. I asked Alexa to book a reservation for me and my wife at a new local sushi restaurant we’ve been meaning to try – and finally could with our kids staying the weekend with a relative. Disappointingly, though, Alexa replied that she couldn’t make a reservation at that restaurant – the restaurant doesn’t use OpenTable for its reservations and that’s the only current partner that Alexa+ has in the space. Alexa instead simply offered me the restaurant’s phone number which….was not exactly what i was looking for. It’s possible that Amazon ends up cutting a deal with Resy, the restaurant reservation service that the restaurant in question uses. While Panay said more partnerships were in the works, neither he nor a spokesperson would confirm specifics.
That said, ordering an Uber by voice worked seamlessly (once I agreed to provide access to my Uber account), though I do wonder how often people will opt for this experience versus simply pulling out their phone. Browsing and homing in on the cheapest soccer tickets at a nearby stadium also worked quite well though, again, I wonder if talking out loud to a virtual ticket assistant for 4 minutes is actually any better or more efficient than searching for the tickets on my phone or computer.
Panay told me beta feedback so far is “overwhelmingly positive,” and that the “conversational aspect” of Alexa+ alone—versus the prompt and response mode of the original—is delighting customers. “It’s just a part of the kitchen conversation at this point,” he noted, emphasizing his point with an anecdote about his family settling debates or open questions by querying Alexa+ rather than pulling out a phone and falling prey to all the distractions that come with it.
“It’s the idea of being engaged with each other and having an ambient assistant there, where I’m not turning on my phone, I’m not opening an app, I’m not being distracted by whatever it is that is on my notifications,” he said.
One major caveat is that I wasn’t able to try out everything that Amazon is excited about. Panay stressed that while engagement with “traditional features” like playing music are increasing, household-management capabilities of Alexa+ are a hit with early users and he believes they’ll continue to be. In one example, he discussed giving Alexa access to a family’s calendar and then prompting it for the best weekend to get away. I haven’t tried that feature mainly because you can’t yet link work email accounts from Google or Microsoft to Alexa+, and because our kids’ sports calendars are spread across several apps that I’m frankly too lazy to consolidate (yes, embarrassing).
“Please don’t underestimate the power of this”
Amazon’s head of devices Panos Panay at the Alexa+ launch event in February 2025
Andrej Sokolow/picture alliance via Getty Images
Panay also highlighted shopping tools powered by Alexa+ that notify you when a certain product goes on sale. And he stressed the ease with which Alexa users who have outfitted their home with smart devices—think smart lights and smart locks —will be able to speak into existence complex routines.
“Alexa, every night at 8:30, start dimming the lights in the house and then lock the doors,” he said by way of example.
That’s four separate commands in one sentence, versus what would have taken at least a dozen and a half steps within the Alexa app previously, Panos said.
“Jason, please don’t underestimate the power of this,” Panay urged me.
One approach Amazon and Panay could take would be to set expectations a bit low and then overdeliver after such a long wait. After all, the introduction of the original Alexa occurred in a really understated way; it was buried within a larger announcement unveiling a surprise device called the Echo.
Panay, for his part, acknowledged that there is still work to do before the new Alexa is ready to be used by hundreds of millions of existing users. And after such a long wait—with Panay himself setting expectations high—it’s fair to wonder if “pretty good” is anywhere good enough in the new world that Amazon’s famed voice assistant is now reentering. Clearly, there’s more work to do.
Could Walmart become a leader in the burgeoning agentic AI race?
After watching the retail company’s technology leaders discuss a host of new agents Wednesday at a New York City event, a yes might not be as farfetched as it might sound to some.
The retail giant unveiled its vision for how AI agents are going to overhaul how customers shop on its digital platforms; how corporate and store employees do their jobs; and how vendors and sellers track their merchandise performance. In some cases, this autonomous technology is doing so already.
“Walmart is all in on agents,” the company’s chief technology officer, Suresh Kumar, told reporters at the event. “Agents can make life simpler for every aspect of what we do at Walmart,” he added.
Despite its roots as a brick-and-mortar retailer, Walmart has more recently been at the forefront of online commerce. In embracing AI agents, however, the company is positioning itself ahead of even many digital companies.
Agents, to many in the tech industry, are the next evolution in the current AI boom, where artificial intelligence not only acts as an assistant, but can autonomously complete complex multistep actions with limited, or even no, human involvement. And for Walmart, the company’s leaders say it’s a natural next step in a technological transformation that has been underway inside the Arkansas-based retailer for the past few years. Kumar said he believes that Walmart holds a key advantage over many competitors in this space, considering the depth and breadth of data the company holds both because of its massive customer base, and when it comes to employee experiences as the world’s largest nongovernment employer.
He and other Walmart tech leaders showed off examples of four “super agents,” which essentially act as managers that rout tasks to each more specialized agent. For consumers, there’s Sparky, currently a generative AI digital assistant that can answer product questions and make suggestions, and which has been live in Walmart’s app for some time. In the future, the assistant will start to take actions. Namely, create an order of weekly essential products based on a customer’s shopping behavior, and place the order with essentially a thumbs-up from the customer. The agent will also eventually possess the capability to curate a multi-item order geared to an upcoming party or event—based on specifics such as theme, attendee size, and a shopper’s budget.
Other leaders showcased internal agent use cases that the company says will more efficiently accomplish mundane and repetitive tasks for store workers, corporate staff, Walmart software engineers, and brands and other companies that sell through Walmart’s physical and digital storefronts.
While some of these agentic use cases are live today, others are coming soon, company execs said. But they were intent on making one point clear.
“It’s not vaporware,” one executive said, accurately reading between the lines of one of this reporter’s questions.
Critically, many questions remain unanswered. What exact impact will this so-called agentic future—if brought to full fruition—have on employee headcount at the world’s largest nongovernment employer?
“We expect jobs to evolve, and we don’t know what that looks like yet,” Walmart exec Dave Glick told Fortune.
Will the revenue and employee productivity gains outweigh the intense costs of using AI at scale, especially for a company known on Wall Street for consistently generating profits?
And at a broader industry level, is Walmart willing to participate in a possible future where consumers trust shopping agents from companies like OpenAI or Perplexity to autonomously make purchase decisions for them? Walmart U.S. CTO Hari Vasudev told Fortune that the company is building the technological capabilities to do so, but that the ultimate decision will lie elsewhere in the company.
“I don’t want to mandate the business model; I want to be able to build it as open as I can,” he said. “Whether the business decides to do it with a particular AI operator or not will depend on the economics and the business model and the relationships.”
Are you a current or former Walmart employee with thoughts on this topic or a tip to share? Contact Jason Del Rey at [email protected], [email protected], or through messaging apps Signal or WhatsApp at 917-655-4267. You can also message him on LinkedIn or at @delrey on X.
I admit it, I have a soft spot for sports cards. Over the past year, an embarrassingly steady stream of small eBay packages have been arriving on my family’s doorstep in New Jersey thanks to my 12-year-old’s newfound love for basketball cards. As I think has become plainly apparent to my wife by now, I’m complicit in encouraging this passion—with more than a tinge of nostalgia driving my mind and heart back to my own baseball card obsession in the late ’80s and early ’90s.
And I’m not alone. Trading cards like Pokémon and sports cards have exploded in popularity in recent years, fueled in part by nostalgic yearning during the early days of the pandemic and new digital gamification elements of the hobby that have turned collecting for some into an FOMO-induced, impulse-buying sport.
Now, one of the fastest-growing startups in the space—New York City-based Courtyard—has raised a $30 million Series A to double down on growth. Forerunner Ventures is leading the round, joined by the company’s existing investors NEA and Y Combinator.
Founded in 2021 by Nicolas le Jeune, who previously worked at YouTube, and Paulin Andurand, a former Apple software engineer, Courtyard markets itself through its website as selling “mystery packs” of cards and comic books via a digital vending machine. What that means in practice is that customers agree to pay either $25, $50, or even $100 for an unknown Pokemon or sports card—or $200 in the case of a comic book—and then an algorithm randomly assigns them a card or comic from the startup’s massive inventory of collectibles stored in the company’s secure vault. Before purchasing, customers can view the probability that they will “pull,” in industry speak, a card of certain value.
If you’re disappointed by the card you receive, you have a few options. Courtyard will immediately purchase the card back for 90% of its fair market value. Customers can also choose to list the item for sale on Courtyard’s marketplace, which doesn’t charge any fees to sellers. That ability to quickly sell out of a purchase you aren’t interested in, or disappointed by, is at the heart of Forerunner’s attraction to the startup.
“Courtyard stands out as the first collectibles marketplace that’s actually designed to be liquid,” Forerunner partner Nicole Johnson wrote in an email to Fortune. “That might sound like a small thing, but it’s a big unlock: it lowers the barrier to entry [for consumers] in a category that’s historically been tough to navigate.”
If the card’s a keeper, customers can keep it secured for free in Courtyard’s storage facility, and have it shipped to them. (Many customers don’t realize, and don’t really need to, that they are in reality purchasing an NFT, or digital token, that represents a specific physical card—or piece of merchandise.)
The mystery packs have been a hit. In January of 2024, Courtyard was selling about $50,000 of merchandise a month. Today—just a year and a half later—le Jeune says the company is selling $50 million a month. And that’s with the comic book category just recently launching and with all sales happening via a website; the startup is expected to release its first mobile app in the coming days. The startup makes money when it buys cards back from customers for 90% of its value and resells it to customers in a new mystery pack. The same card is sold an average of eight times a month on the platform, the company said. Courtyard also relies on a large network of collectibles dealers who source merchandise for the startup, helping make the startup the largest buyer of trading cards in the world right now, according to le Jeune.
But clones are popping up and the CEO said he is willing to go into the red to step on the gas and expand the startup’s lead, through a mix of hiring, paid marketing, and product category expansion.
“We want to make sure we double down on the growth and capture the market as fast as possible,” the CEO said.
When I tried out the service recently, I purchased a $25 mystery basketball card. It turned out to be a rookie card for Jalen Green, a young NBA player drafted as the 2nd pick in the 2021 NBA draft by the Houston Rockets. Green was recently traded by the Houston Rockets to the Phoenix Suns and has not lived up to his initial promise. Courtyard offered me $9 for a card valued in the broader basketball card market at $10. I didn’t accept, and have chosen to store the card with Courtyard for now, hoping that Green might play better this year and the value of the card might increase, making it worth it to keep or to add to my son’s collection. For a first experience, though, it was a bit disappointing even if I knew logically that pulling a card of greater value was not at all guaranteed.
I asked le Jeune about the risk that a first-time buyer experience like mine might dissuade a customer from sticking around and making another purchase.
“We could fake it and make you feel at the beginning like you get a good card every single time,” he said, “but we would not be okay with that.”
As the company gets bigger, he added, Courtyard will be able to offer more value to its customers for each transaction.
“We grew so fast but it’s still the early days,” le Jeune said, “and there’s so much room to make it a much better experience.”
Login
