The Little Known Chip Lab Behind Amazon’s Success
13:49 Share

In case you missed it, YouTube is the number one streaming platform in watch time in the U.S., ahead of Netflix, Disney, and Prime Video for the second year in a row. There's only one YouTube.

Hey, TNB listeners. Before we get started, heads up. We're going to be asking you a question at the top of each show for the next few weeks. Our goal here at Tech News Briefing is to keep you updated with the latest headlines and trends on all things tech. Now, we want to know more about you, what you like about the show, and what more you'd like to hear from us.

What kind of stories about tech do you want to hear more of? Business decision-making? Boardroom drama? Peeking inside tech leaders' lives? Tech policy? If you're listening on Spotify, look for our poll under the episode description. Or you can send an email to tnb at wsj.com. Now, onto the show. Welcome to Tech News Briefing. It's Monday, May 12th. I'm Julie Chang for The Wall Street Journal.

Apple is facing an existential crisis. Should it keep focusing on hardware, what it's been known for, or pour its energy into improving software, namely Siri? Our senior personal tech columnist will tell us what the two paths would look like for the iPhone maker. Plus, we'll tell you about the secretive chip lab cooking up Amazon's secret sauce to success.

But first, it's been a tough year for Apple. Its artificial intelligence capabilities aren't living up to the hype, it's navigating tariffs, and it's being investigated over antitrust violations, among other things. WSJ senior personal tech columnist Joanna Stern says, of all those issues, AI might pose the company's biggest long-term threat. She's with me now. Joanna, explain that for us. Why is AI such an issue for Apple?

Well, if you think about AI right now as sort of this new underpinning technology that's powering pretty much all new software applications and operating systems, Apple has to integrate AI throughout its products. And Apple's always been known for making great hardware, but also making software that goes with it. So that software is all about AI now. And Apple's got to keep up there to

really start to rival what their competitors like Google, Microsoft, OpenAI are doing. - And you think that there are two paths when it comes to Apple and AI. Can you tell us about the first path? - We cannot talk about Apple and AI without talking about Siri, this promise of an AI assistant that is gonna do everything for us. And that's been the promise since 2011.

And Apple's been trying to live up to that promise. So path one is that Apple completely fails at AI. And my funny name for this is called misery or miss Siri, because we're all going to live in this world where we're just miserable using Siri. So what happens if Apple doesn't get Siri right? Would that have an impact on Apple hardware products?

So presumably Apple continues to make great hardware. But what happens in this case is that the other types of AI providers, the ones that we're starting to really lean on now, ChatGPT, Perplexity, start to take over more of the software interface. So when we go to use our phone, we're not talking to Siri. We're not using Apple's apps. We'd end up with the iPhone and many of the other devices that Apple makes just being a vessel for other companies. There's one more big fear, and that's that other companies become really good at hardware.

And we're already starting to see this happen. We're seeing it with Meta. I'm a huge fan of these Meta Ray-Ban glasses. They're simple glasses. They have cameras in it. They have microphones and speakers. And you can use it to talk to Meta's AI. And it's really helpful. And so if other companies start making this hardware really well and integrated with their AI, well, people may start to buy that instead of Apple.

So now take us down the second path that you envisioned here. What happens if Apple nails AI? I call this Siri-topia. Siri works as we've always been promised. It is very much like we're starting to see with these chat GPT voice assistants online.

or Meta's voice assistants, where it sounds more human, it can hold a conversation, it does what you asked. Siri is far more responsive. And what that leads to is the opposite of what would happen in Path 1, right? Which is that we've got iPhones, we've got AirPods, we've got Apple Watches, we've got maybe glasses that Apple is reportedly working on, that all work really well. And Siri's the underpinning assistant, the AI that's powering it all.

Has Apple said anything about its AI path? Apple declined to comment for the story, but I will say something really interesting happened. As I was putting this story to bed, Eddie Q, who's Apple's senior vice president of services, was testifying in the Alphabet or Google antitrust case on Wednesday. And he was asked a lot about Apple's plans in AI as it relates to Google. And he said a couple of things. I'll read one of the quotes because I think it really summarizes what Apple is up against here.

We're highly successful. That doesn't mean we're going to be around 10 or 20 years down the line. You have to earn it in technology every day. We're not an oil company. We're not a toothpaste. These are things that are going to last forever. People are going to need toothpaste 20 years from now, 40 years from now. You may not need an iPhone 10 years from now, as crazy as that sounds. And so I think that is just a perfect example.

encapsulation of how Apple is viewing AI as an existential threat to its entire business, including the hardware. And Apple's Worldwide Developers Conference is coming up. What are you going to be watching for? Apple had a really tough year. Last year at WWDC in June, executives talked about AI and new improvements coming to Siri. Apple announced that those were going to be delayed.

And so this year, we are looking to Apple to not only talk about what they're going to do in AI, but also are they going to deliver on it? That was WSJ senior personal tech columnist Joanna Stern. Coming up, how a little-known startup is playing a big role in Amazon's AI strategy. That's after the break. Did you know that every day people watch on average more than 1 billion hours of YouTube on their TV screens?

That's because YouTube is where people go deep on all the content they love. There's only one YouTube. Annapurna Labs may not be a household name, but it's a company that's helped Amazon shoot up in success through an acquisition long ago. The chip designer essentially underpins the e-commerce giant's AI strategy. Its custom silicon has even been described as the secret sauce of AWS or Amazon Web Services.

Ben Cohen wrote about this for the Wall Street Journal's Science of Success column. Ben, tell us about Annapurna. Annapurna Labs is this Israeli chip design company that started in 2011 in Israel. And in 2013, it got into business with Amazon. And in 2015, Amazon bought the company outright for reportedly about $350 million. And

That was 10 years ago. It seemed like a lot of money at the time, and now it seems like a bargain because it turned out to be one of the most consequential deals in the history of enterprise technology. And this one team within the AWS division of Amazon has become essential to the success of the entire company.

So before we dive into the specifics of what Annapurna does, can you tell us a little bit about how this relationship with Amazon started? Yeah, it started with this clandestine meeting in a bar in Seattle's Pike Place Market called Virginia Inn. And Nafa Bashara, one of the co-founders of Annapurna, was there to meet James Hamilton from Amazon. James Hamilton is an executive at Amazon. His official title is Distinguished Engineer. And

So Nafa Bashara is from Israel, but he's based in Silicon Valley, and he flew up to Seattle for this meeting. He thought that it would be awkward to present from his laptop in a bar, so he just brought four slides that explained what Annapurna was, what they did, how they did it, and why they should do it for AWS.

That meeting over beer and wine in the Virginia Inn eventually led to this relationship between Annapurna and AWS that led to this deal. That is really what started all of this. And all of this is incredibly essential to Amazon these days as it tries to keep up in this trillion dollar AI arms race.

Yeah, let's dive into that a little bit. So you write that Amazon Web Services specifically is very dependent on Annapurna. What exactly is it that Annapurna provides for them?

So Annapurna designs and researches all of the chips that are essentially the foundation of the company's entire AI strategy. So Andy Jassy, the CEO of Amazon, who was the CEO of AWS when they made this deal for Annapurna, has explained how there are certain layers to

what Amazon is trying to do with AI. And the bottom layer is made up of chips. So some of the chips that are available in AWS data centers are designed by other companies like Intel and NVIDIA. And some of them are designed by Amazon. And those chips that are designed by Amazon, they come out of Annapurna Labs. And so the big one right now is this chip for training AI models called Trainium.

Annapurna also designed a CPU chip for AWS called Graviton that was like their first big success, their first big breakthrough chip for Amazon. And all of this is meant to improve price performance for AWS customers and give them this variety of options depending on what they need at any particular time. Like that diversity factor

of choice is really important to the AWS strategy. And it's almost like they're offering them a buffet of chips. Some of them are homemade, some of them are from others, but they want AWS customers to be able to use whatever they want on the AWS cloud, almost like the Amazon store itself. Yeah. So when I hear Amazon, I think amazon.com. So how big is the AWS division compared to the rest of the company?

It's really, really big. AWS itself brings in $100 billion of revenue for Amazon now. And more important, it brings in more than half of the entire company's profits. That's like basically what Target brought in revenue last year. It's as big as some of America's biggest companies. Amazon has long depended on Amazon Web Services, and increasingly Amazon Web Services depends on Annapurna Labs. Looking forward, where do Amazon's AI ambitions lie? And what role is Annapurna playing in that?

Amazon, Google, and Microsoft are all spending just gigantic sums of money to build out their AI infrastructure right now. And no company on the planet is spending as much on CapEx at this point as Amazon, which is planning $100 billion of investments this year, mostly on AI for AWS data centers. And they're all trying to keep up.

which means spending more and more on chips, in part because they want to take greater control of their own supply chains, in part because it lowers their reliance on NVIDIA, which still dominates the market for AI. And that's sort of where Annapurna comes in. They have designed this chip for training AI models called Trainium. And

Still, NVIDIA chips handle the vast majority of workloads in AWS data centers. And the question is, can Tranium chip away at that lead? And if they can, what does that mean for Amazon? And it's the same reason that

Google and Microsoft are developing their own custom silicon. They all want more control in this process. They all have a steep climb ahead. And if you ask Andy Jassy, the CEO of Amazon, he'll say it's still very early days. And even though it's been 10 years since the Annapurna acquisition, it'll be more interesting to discuss all of this 10 years from now. That was Science of Success columnist Ben Cohen.

And that's it for Tech News Briefing. Today's show was produced by Charlie Duffield with supervising producer Melanie Roy. I'm Julie Chang for The Wall Street Journal. We'll be back this afternoon with TNB Tech Minute. Thanks for listening. The world's biggest creators, the world's biggest moments, all delivered to the world's biggest collection of passionate fans, providing unparalleled opportunities for your brand. There's only one YouTube.