AWS to Inject $50B Into Federal AI Compute Reliability

0:00AWS is spending$50 billion to build AI infrastructure for the US government. Today on the podcast, I want to break down why they're able to do this, how this deal was put together, and why I think this is a bigger deal, not just for the US government, but for all governments around the world who are going to be doing similar deals to help create sovereign AI, why creating an AI advantage is important for most countries. We're going to be getting into all of that. Before we do, I wanted to mention, if you want to try any of the models that I talk about on the show, I'd love for you to check out my startup, which is called AIbox.ai.

0:33You get access to 40 of the top models, OpenAI, Claw, Gemini, Grok, 11 labs for audio, tons of cool image models, all for$20 a month in one place. You can chat with them all in the same thread. It makes it super useful, super easy, and very affordable. So if you want to check that out, go check out AIbox.ai. I'll leave a link in the description. Okay. So AWS is making a huge new investment in infrastructure. Basically, this is designed to boost the United States government's AI capabilities. They made this big announcement. They're going to put$50 billion into this. And what they're saying is that this is quote unquote high performance computing infrastructure.

1:09This is going to be built for the U S and the build out is meant to expand the federal government agencies access to AWS's AI services. Now that is there, you know, there's a lot of different agencies within the federal government. I think they're kind of doing this big deal and different agencies inside of the government are going to be able to then have access to this. It's going to add about 1.3 gigawatts of compute and it is going to expand the US government's access to AWS products. There's a bunch of different things that will be bundled into this. There's Amazon SageMaker AI, there's model customization, there's Amazon Bedrock, there's model deployment, and of course, Anthropics Cloud Chatbot, which is very deeply integrated with Amazon after they've invested over$4 billion in the company.

1:52So it's kind of interesting because it's beyond just the compute and beyond just building out the data centers, it's also getting something like Anthropics Cloud into the US government. So AWS is expecting to start working on all of these projects in 2026 next year. And this is what they said about this whole situation. This is the CEO Matt Garman of AWS. In a recent release, he said, our investment in purpose-built government AI and cloud infrastructure will fundamentally transform how federal agencies leverage supercomputing. We're giving agencies expanded access to advanced AI capabilities that will enable them to accelerate critical missions from cybersecurity to drug discovery.

2:30This investment removes the technology barriers that have held government back and further positions America to lead in the AI era. Now, I think it's kind of interesting because like on the one hand, they're like, we are like, you know, make unlocking AI for the government. And the government already is, the US government already has contracts and deals made with OpenAI and Cloud and Anthropic and like a lot of these other companies. But I do think this is a big deal when the government is actually getting access to the data centers and compute. They can fine tune and train their own models. There's a lot of competitive advantages that they'll be able to do that, you know, these models don't do right out of the box.

3:06And of course, the US government and probably every government has many unique data sets that no one else has that they can train data off of. they can train models off of and their exclusive data sets that no one else is ever going to get access to. So I think there's a lot of interesting, like technically, I hate to say like the government's innovative, but there's a lot of interesting innovations and maybe some people are terrified about those innovations. Some of them obviously are good for national security and I'm sure other people will find some of them questionable for other reasons. But I mean, that's a whole nother can of worms.

3:37I think the entity AWS started building the cloud infrastructure for the government back in 2011. So this is not a new deal by any means, but this is a really big step and a really big use case. Three years later, they launched AWS's top secret East. It was the first air-gapped commercial cloud to work with classified workloads. AWS introduced AWS secret region in 2017, which has accredited access to all levels of security classifications. So they've been building a lot of this custom stuff for the US government for a very long time. This is not anything, you know, incredibly new, but in an age where AI is out and these AI tools make massive productivity gains, we've seen inside of the enterprise, you know, the U S government is kind of banned, not really banned, but they're blocked from using a lot of these tools because of the security and a lot of other, other, you know, there's top secret confidential things that we can't be put into these models.

4:30And so building these data centers where they can run their own models, they can build these own in-house integrations really unlocks a lot for them. So I think obviously we've seen a lot of different tech giants pitching AI services to the US government over the last year, I think, is when this has been getting really popular. OpenAI launched a version of ChatGPT designed exclusively for federal US government agencies back in January. And OpenAI announced a deal in August that essentially gave government agencies access to the enterprise tier of chat GPT for just$1 a year. So basically they gave it away for free.

5:03Some people could say, well, they're, you know, they're bribing the government to try to get favorable, you know, to be viewed favorably by different enterprises. What's kind of interesting is like if OpenAI, for example, is like theoretically, let's say it was going to get investigated for anti-competitive, you know, use or, you know, for being anti-competitive by the FTC or by the Federal Trade Commission or whoever, right? You had Lena Kahn back in the day who was reigning with an iron fist and is no longer doing that. So it seems like a lot of the tech is not being quite as heavy handed in the regulation.

5:40But let's say that they were being investigated for one of these issues. Well, the people that are investigating them will now have access to OpenAI, to ChatGPT, to do their investigation on OpenAI and ChatGPT, will that curry favor with the people doing the investigation being like, well, this is pretty useful. I don't really want to ruin this company. Is that their motive for doing it or was their motive just trying to show off to the American people to say, look, we're good. We're giving stuff away for free. Was it for the leaders of the organizations? Maybe Trump is going to see them giving this away and really like them.

6:12In my opinion, I actually think all of those cases would be great reasons for OpenAI to give the government$1 a year access to chat GPT. They're definitely not losing and they have a lot to gain in the situation and not that much to lose. So the same month that they did that, Anthropik also announced it was giving the US government access to the enterprise tier of Claude chatbot for$1. Google announced Google for government for even less charging 47 cents for the first year. So I think everyone was kind of like trying to one up each other being like, no, no, no, no, like we're, you know, we're the guys that you want to not regulate and destroy.

6:47And also like a lot of federal, the federal workforce is a big, you know, a big group of people that is usually highly educated and has good salaries. So like this is a quite a desirable workforce that was blocked from using these tools for a lot of different reasons, unless they kind of have these special programs. So it makes sense. AWS is not the first person to get deeply integrated and build these things out. We have OpenAI, Google, Claude, everyone seems to be getting in on it right now. So we'll see how that rolls out in the future and if that makes much of an impact on regulation of these companies or anything else in that regard.

7:19Thank you so much for tuning into the podcast today. If you enjoyed the episode, make sure to leave a rating review wherever you get your podcasts. It helps us show out tremendously. And as always, make sure to go check out AIbox.ai if you want to get access to Google, Claude, OpenAI, 40 different top models for audio, text, and images, all for 20 bucks a month. It's linked in the description and hope you have a fantastic rest of your day.