Multibillion Dollar AI Infrastructure Build Out Coming to the US

0:00Today on the AI Daily Brief, a massive AI infrastructure spending plan comes into view. Before that, in the headlines, the UK says there's no problem with Microsoft's deal with inflection. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show notes.

0:23Welcome back to the AI Daily Brief Headlines Edition, all the daily AI news you need in around five minutes. We kick off today with a follow-up on a story we've been watching for a while, which is antitrust scrutiny around Microsoft's sort of but not really acquisition of inflection. This weird new thing has been a big trend in AI for the last year or so, where a big tech firm, Microsoft, Amazon, and Google have all done this recently, hires the founders of a big startup, licenses the technology, and then invites some but not all of the employees to come over and join. Now, in the case of Microsoft's sort of but not really acquisition of inflection, a big chunk of the team really did come over.

1:00The way that the deal was structured, it paid back investors, but it's not like they got a big return or anything. We've recently seen other deals like this, most notably Google's recent deal with Character AI. And as I've discussed previously, I think that these types of deals have fairly significant implications for, if nothing else, the enthusiasm with which people will sign up to work with AI startups. If the norm becomes that startups are simply a junket for founders to get a big payday, but no one else makes money, it's not really going to be a sustainable situation. Now, the speculation has, of course, been that a big part of the motivation for this is to avoid antitrust scrutiny, that particularly in the area of AI, which already has so many factors making it more concentrated, the big tech firms are worried about picking off smaller competitors, or particularly what governments will think if they hire those smaller competitors.

1:44The UK had been looking into Microsoft's specific deal with inflection in an examination that started back in July. The UK's Competition and Markets Authority ultimately decided that even Even before it was acquired, Inflection had only a very small portion of the UK chatbot market and didn't really have a meaningful chance to be a competitive player long term in that space. The conclusion then was that the deal did not require a deeper investigation. Of course, the question becomes, if that's the conclusion that someone like the CMA came to, wouldn't that apply if Microsoft had acquired the company outright?

2:14Now, ultimately, of course, Inflection could have just made a decision that this was the best deal that they were going to get. And it might be a situation where it's very easy to armchair quarterback it from the outside. But all in all, good news for Microsoft, who obviously didn't want to get stuck in legal mire when it comes to this deal, which really restructured their AI team in general. Microsoft, it seems, has enough to be dealing with when it comes to AI. A recent report from the Information suggests that the Copilot software in the Office suite has had something of a challenging time getting customers excited.

2:41The Information writes that, quote, So far, only a small portion of Microsoft's business customers have adopted Copilot. Between 0.1 % and 1 % of the 440 million existing users of Microsoft 365 are also paying for the new AI features. Now, notably, that's an equity analyst estimate. That's not direct numbers from Microsoft. Still, that translates to somewhere between 400 ,000 and 4 million paid seats. Competitors are swarming to encourage the narrative, it seems as well. On a quarterly earnings call last week, Salesforce CEO Mark Benioff said, So many customers are disappointed about what they bought from Microsoft co-pilots because they're not getting the accuracy and the response that they want.

3:14Microsoft has disappointed so many customers with AI. The silver lining is that the companies that the information spoke to seem to be in it for the long haul. One chief information officer that they spoke to, for example, said, So far we've had mixed results. Most people don't find it that valuable right now, but it's a product that's going to improve over time. Staying in the UK for a moment, the US, the European Union, and the UK have signed the world's first legally binding international AI treaty. It's the Council of Europe's AI Convention and makes signatories accountable for harms and discrimination that come from AI systems.

3:45It mandates that the outputs of AI systems respect citizens' equality and privacy rights and gives the victims of AI rights violations legal resources. Specific consequences, however, are yet to be put in place. Now, although the treaty is legally binding, it certainly seems like something that is largely symbolic rather than legally significant. If anything, I think it reflects the continuing interest that the world has in trying to create some sort of regulatory systems around this new field. Moving into the enterprise, Anthropic has announced Cloud Enterprise, which they call its biggest new product since Cloud debuted in the first place.

4:16One of the big updates is that Cloud Enterprise has a much bigger context window than previously. It'll allow clients to upload documents that are the equivalent of 130-minute sales conversations, 100 ,000 lines of code, or 15 full financial reports. Anthropic has been on a tear recently, pushing out new features including artifacts, as well as new workspace tools. Salesforce has made an acquisition in the AI space, acquiring AI voice agent developer Tenix, which they say will, quote, extend Salesforce's existing autonomous agent capabilities for a service agent by integrating Tenix's innovative voice AI solution specifically tailored for service use cases.

4:49TLDR, the enterprise world is continuing to move aggressively towards an AI agent paradigm, and that's happening both top-down from big companies as well as bottom-up from startups. Speaking of, a cool startup that I've been noticing recently that just hit the news waves is a company called Paradigm, which is making a bet on agents to change the way that we use spreadsheets. There are a bunch of really interesting projects in this space, and it's something that we're watching closely. For now though, that is going to do it for today's AI Daily Brief Headlines Edition. Next up, the main episode.

5:17Today's episode is brought to you by Plum. Generative AI promises to supercharge your productivity and give you superpowers, but if you're not an engineer, trying to harness AI can be incredibly frustrating. Hours wasted wrestling with complex tools only to give up when they don't work. We all have tedious tasks we'd love to automate and challenges AI could solve, but few of us have the skills to fully leverage these game-changing technologies. That's where Plum comes in. The mission? To make automating your work feel like magic. Imagine typing out, AI, read my Gmail and ping me in Slack when something critical comes in, and watching it come to life before your eyes.

5:50No coding required. Whether you're a marketer, salesperson, or founder, Plum enables you to create custom AI workflows in minutes, not hours. Check out useplum.com, that's Plum with a B, for early access to the future of workflow automation. Today's episode is brought to you by Plum. Want to use AI to automate your work but don't know where to start? Plum lets you create AI workflows by simply describing what you want. No coding or API keys required. Imagine typing out, AI, analyze my Zoom meetings and send me your insights in Notion, and watching it come to life before your eyes. Whether you're an operations leader, marketer, or even a non-technical founder, Plum gives you the power of AI without the technical hassle.

6:24Get instant access to top models like GPT-40, Cloud Sonnet 3.5, Assembly AI, and many more. Don't let technology hold you back. Check out Use Plum, that's Plum with a B, for early access to the future of workflow automation.

7:28Today's episode is brought to you by Super Intelligent, which is, of course, our platform that helps you learn how to use AI tools and perhaps even more importantly, gives you ideas on the best use cases that are actually going to help you achieve whatever it is you want to achieve. To recognize the end of summer and back to school slash back to work, we are running our best promotion ever. When you sign up for Superintelligent, between now and the end of August, using code SOBAC, your first month will be 100 % free. The platform features over 600 fun, highly practical AI tutorials that get you using AI fast and with an eye to actually transforming how you get things done.

8:08We've just launched Super for Teams, so if you have a group of people at your company that want to figure out how to use AI together, I highly suggest you check it out. But for those of you who are using Superintelligent as an individual, once again, if you sign up for Superintelligent between now and the end of the month using code SOBAC, you will get your first month 100 % free. Go to besuper.ai and check it out today. Welcome back to the AI Daily Brief. Today, we are talking about a big plan to build out AI infrastructure in the US. And to get the proper background, we have to go back a little bit, as we so often seem to do, to when Sam Altman was fired by OpenAI, before, of course, being quickly rehired.

8:47One of the reasons that was cited for Altman's dismissal was concerned that he was doing things that weren't related strictly to OpenAI, leveraging OpenAI's name. A couple months later, reports started coming out that Altman was gallivanting around the world seeking a huge amount of money. We're talking trillions of dollars to build out AI infrastructure. In February, the Wall Street Journal reported that, quote, the OpenAI CEO was in talks with investors, including the United Arab Emirates government, to raise funds for a wildly ambitious tech initiative that would boost the world's chip building capacity, expand its ability to power AI, and cost several trillion dollars.

9:21The project could require raising as much as$5 trillion to$7 trillion. Now, for some sense of scale, the current size of the global semiconductor industry, in 2023, chip sales reached around$527 billion, and that's expected to rise to a trillion annually by 2030. Semiconductor manufacturing equipment is a$100 billion a year industry. As the WSJ points out, the quote amounts Altman had discussed would also be outlandishly large by the standards of corporate fundraising, larger than the national debt of some major global economies and bigger than giant sovereign wealth funds. So this was what was going on back in February.

9:54Well, now we're getting a little bit more information. Bloomberg reported this week, Altman infrastructure plan aims to spend tens of billions in the U.S. Global plan for AI would start with a major push in the U.S. Bloomberg writes, Altman spent the early part of the year seeking the U.S. government's blessing for the project, which aims to form a coalition of global investors to fund the costly physical infrastructure needed to support rapid AI development. Now, Altman and his team are working on several details, including the plan to first target U.S. states. Bloomberg continues, The types of projects under discussion include building data centers, increasing energy capacity and transmission with turbines and generators, as well as expanding semiconductor manufacturing.

10:30Now, the Bloomberg piece makes it seem like this is not strictly separated from OpenAI. They write that OpenAI executives, including Altman, but not exclusively Altman, have been meeting investors in recent weeks to advance the deal. OpenAI's chief strategy officer, Jason Kwan, for example, recently traveled to Japan and Korea to speak about the plan. Their VP of global policy, Chris Lehane, spoke to investors in Canada. This all seems to suggest that this is in fact an OpenAI initiative. Now this is all reports and not confirmed, but when asked about the plan, OpenAI didn't exactly issue a denial.

11:01A spokesperson said that the company, quote, believes building additional infrastructure in the US is critical to further advancing AI and making its benefits widely accessible. We are exploring opportunities with this goal in mind and look forward to sharing more details at a later date. The reporting also tells a little bit more about the tie-up with the US national security infrastructure. OpenAI has apparently held meetings with the U.S. National Security Council about the investment, in which, according to Bloomberg sources, Altman and other OpenAI executives emphasized that the global infrastructure plan would help fuel America's geopolitical advantage by establishing a multinational coalition to compete with China's own infrastructure consortium, and that foreign capital that might otherwise go to China would go to the U.S.

11:38One of the subtexts of this, then, is a competition to win the Middle East, specifically the Gulf states, to the U.S.'s AI cause rather than to China's. Overall, I think that while it's early, this could be an incredibly significant bit of news for the way that AI develops, and so I am going to keep a close eye on it. In the meantime, it's clear that AI infrastructure continues to be a big concern. The information recently reported that two different AI developers are planning$125 billion supercomputers. The information reports that, quote, recent comments from a state official in North Dakota signaled that some companies are exploring deals that would enable proposed new data centers to become real.

12:12During a recent Public Service Commission meeting held about data center demand, the North Dakota Commissioner of Commerce said that two companies had approached him and the governor about building mega AI data centers. The projects would both consume around 500 to 1000 megawatts of power with plans to scale up to 5 to 10 gigawatts of power over several years, making them then orders a magnitude bigger than any data centers that are in existence today. The information writes, to put the scale of the plan in perspective, at the end of last year, Microsoft's global data centers for its Azure cloud computing business consumed around 5 gigawatts of power combined.

12:43While the specific companies were not named, the official said that the companies have, quote, trillion-dollar market capitalizations, which basically means that they're talking about NVIDIA, Amazon, Microsoft, Google, Meta, or Apple, perhaps Tesla, which has been a trillion-dollar company in the past, although is now around$700 billion in market cap. Now, in terms of how likely to see something materialize we are, the official said in that meeting that, quote, things can happen pretty fast when you've got a trillion dollars. That official also seems to be a big proponent, saying, if we don't do it, Texas will do it, Oklahoma will do it, other states will do it.

13:14Speaking of that, Elon Musk's XAI has announced that their Colossus AI training system featuring 100 ,000 GPUs has come online. Musk said from start to finish it was done in 122 days. Colossus is the most powerful AI training system in the world. Moreover, he said it will double in size to 200 ,000 H100 equivalents in a few months. The information noted that the announcement might carry some concern for other AI rivals. They write, if it's true that it only took four months to complete this cluster, its quote rapid buildout would fly in the face of conventional wisdom about how long it takes to build such a system.

13:48Normally completing such a cluster could take a year. Of course, they point out that it would be both on brand for Musk to actually have achieved some nearly impossible goal, but also as they put it that he has a penchant for exaggerations. So was it real? Only time will tell. But I think that the through line for today's stories is that this infrastructure buildout is hugely significant and does not show any signs of slowing down. That's going to do it for today's AI Daily Brief. Until next time, peace.