Amazon (AMZN) has invested another $2.75 billion into generative AI startup Anthropic. It completes a $4 billion total investment the tech giant announced in the fall.
AWS Vice President of AI Products Matt Wood calls Anthropic's Claude 3 "the best performing generative AI model." He says Claude 3l is available to AWS customers, who can build on top of the model to build their needed AI functions.
Wood tells Yahoo Finance's Dan Howley he has never seen such excitement from customers over a new technology. "The opportunity for them is really to be able to just completely reimagine core elements of their business, core elements of their customer experience, and to design and deliver entirely new categories of products," Wood says, noting that there is a lot of enthusiasm from industries that tend to be more highly regulated.
Watch the video above to hear what Wood says the number one thing customers want from generative AI.
For more expert insight and the latest market action, click here to watch this full episode of Yahoo Finance Live.
Editor's note: This article was written by Stephanie Mikulich.
Video Transcript
SEANA SMITH: Well, Amazon is cutting hundreds of jobs in its AWS Cloud Division, mostly affecting sales and marketing employees, as well as the team developing their, quote, "Just Walk Out" store technology. Now, this comes on the heels of the company announcing that they are going to invest another $2.75 billion in the AI research and security company Anthropic. Yahoo Finance's tech reporter Dan Howley has more on that for us. Dan.
DAN HOWLEY: That's right, Seana. We're here with Amazon AI VP Matt Wood. He's here to talk to us about, what else, generative AI and Anthropic, obviously. We have the Anthropic and AWS logos back here. So, Matt, what can you tell us about the Anthropic deal? I know that you-- Amazon completed the $4 billion investment. What does that do for Amazon and AWS?
MATT WOOD: Sure. Thanks for having me.
DAN HOWLEY: Sure.
MATT WOOD: Yeah, we're a big fan of Anthropic. We just, as you say, just completed our $4 billion investment. It's a equity investment, convertible note. We don't have a role on the board or anything like that. But Anthropic make generative AI, what they call foundation models. And that's really the central robotic core for the generative AI revolution. Anthropic today have the best performing generative AI model. It's called Claude 3. We made that available on AWS. And customers build on top of that model to do document management and drive automation and digital transformation inside their organization.
DAN HOWLEY: So what is-- what are you seeing from customers as far as uptake when it comes to generative AI? Are they clamoring for it? Are they tepidly stepping into it? Are they, you know, going full bore?
MATT WOOD: There is nothing tepid about generative AI today. I have-- honestly, I've been at Amazon now for nearly 15 years, and I have never seen such enthusiasm, momentum, and investment from our customers around a technology. And the opportunity for them is really to be able to just completely reimagine core elements of their business, core elements of their customer experience, and to design and deliver entirely new categories of products. And we're seeing that just across the board in so many different industries.
Of interest, there's just a huge amount of momentum in the regulated industries, industries that maybe don't always have the best reputation for investing in new technologies-- health care, insurance, financial services, life sciences. They are all over generative AI and using it to develop new approaches, everything from chatbots all the way through to Pfizer delivering new cancer targets.
DAN HOWLEY: So what are they asking for when they come to AWS and they say, we're interested in generative AI? What-- what's that process? And is there anything in particular that every company is saying, we want that?
MATT WOOD: Yeah, I mean, the number one thing, do not pass go, do not collect $200, is privacy and security. And what we've heard from customers is that they don't want to have to make a choice when it comes to privacy and security of their data when using generative AI. And our approach at AWS is that customers want nothing more but also nothing less than they get from any other AWS service when it comes to privacy and security.
And that allows them to be able to safely and securely, privately, to be able to use their own data from within their own organization with these foundation models in a way that information is not exfiltrated. It's not used to train the underlying systems. There's no humans reviewing it. With AWS, they have confidence to be able to bring a lot of that data, which is already on AWS, into their generative AI system, which allows them to build more sophisticated automation.
DAN HOWLEY: So in addition to Anthropic, there's tons of generative AI models out there, the foundation models. What does Amazon offer in particular when it comes to AWS? We've heard, you know, Google's got XYZ. They're also an Anthropic investor. Microsoft, obviously, and OpenAI are basically best friends at this point. Where's-- where's AWS? And what are they offering in addition to Anthropic's?
MATT WOOD: Yeah, so our approach is that we have our own Amazon foundation models. They're great for building automation and natural language tasks. But what we've seen is that there's not really one model to rule them all when it comes to generative AI. Customers want a choice of models so they can find the right choice in terms of capability or price or latency and speed. And so we offer a range of different models from different providers, some from Amazon, some from Anthropic.
We just announced yesterday that we're bringing the latest Mistral AI models called Mistral Large to AWS. We have publicly available models from Meta. We work with Stability AI, AI21 Labs. The list goes on. And so it's really all about helping customers in isolation find the right model for their use case. And then the superpower of generative AI is combining those models and having a compounding effect on the intelligence that you can apply to automation.
DAN HOWLEY: Just one last question. The hardware side of things is always very popular to talk about. Nvidia is one of the biggest stocks of the year and last year. What kind of hardware can people expect? And are they able to select between them? Can they say, I want to use Amazon's own chips, I want to use Nvidia chips, I want to use someone else's?
MATT WOOD: Absolutely. Just like we approach different models and having selection and choice, that is just disproportionately important with the chips and the silicon that customers run on as well. So we have our own chips that we build inside Amazon, one that's specifically designed to accelerate machine learning and AI model training, one which is designed to accelerate model inference or making predictions, generating text, those sorts of things.
But customers can also choose the latest Nvidia GPUs and other accelerators from other providers. So having that choice when it's early like this is disproportionately important. But we think that that choice is going to remain important as more and more customers go further down their journey.
DAN HOWLEY: All right. Matt Wood, VP of AI at AWS, thank you so much for joining us.