The AI Daily Brief: Artificial Intelligence News and Analysis - How US Antitrust Investigations Into Nvidia, OpenAI and MSFT Could Make Things Worse Not Better

Episode Date: June 7, 2024

The U.S. government is investigating Nvidia, OpenAI, and Microsoft for potential antitrust violations. This episode explores how these investigations could impact the AI industry and whether they might make things worse instead of better. The arguments from both sides are examined, the potential consequences are discussed, and the broader implications for the AI sector are analyzed. ** Join Superintelligent at https://besuper.ai/ -- Practical, useful, hands on AI education through tutorials and step-by-step how-tos. Use code podcast for 50% off your first month! ** ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://aidailybrief.beehiiv.com/ Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@AIDailyBrief Join the community: bit.ly/aibreakdown

Today on the AI Daily Brief, the U.S. gears up for antitrust investigations on OpenAI, Invidia, and Microsoft. Before that in the headlines, speaking of Invidia, that company is now worth more than $3 trillion and more than Apple. 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. Welcome back to the AI Daily Brief Headlines edition,

all the daily AI headline news you need in around five minutes. invidia's meteoric rise simply will not stop. We've got CEO Jensen Huang signing women's chests over here, and now Nvidia has surged past $3 trillion of market cap to become the second largest public company in the U.S. trailing only Microsoft and now exceeding Apple. They are now only the third company in U.S. history behind Apple and Microsoft who have crossed that $3 trillion threshold.

The company's stock is up nearly 150% this year, after rising 239% last year. Compare that to Apple, which is up just 1.7% on the year. Part of the reason for excitement this week is that NVIDIA announced that they're speeding up the rate at which they roll out improvements. They announced the new platform called Rubin, which is slated to come out in 2026, which will succeed the Blackwell, which was only announced back in March, and was dubbed at the time the world's most powerful chip. Wrote senior equity analyst Angela Zeno, We think NVIDIA is on pace to become the most valuable company given the plethora of ways it can

monetize AI and our belief that it has the largest addressable market expansion opportunity across the tech sector. Now, for people who think that AI is in a bubble, this is doing nothing to help that out. Investor Sam Lesson wrote, the idea that Nvidia is now more valuable than Apple means something is dramatically mispriced. Either Apple's $400 billion in revenue and $125 billion in EBIT is way too cheap, or Nvidia's $60 billion in revenue and $30 in EBIT is too expensive. Not to play devil's advocate to Sam, who, by the way, is a great follow on Twitter. But part of why the markets are pricing these things differently, is that they're looking at the trend lines of growth in AI and just seeing a massive market expansion opportunity for Nvidia. Whereas for a company like Apple, although yes, their revenue

is eight times Nvidia's, they're a company which is struggling with market saturation and trying to find new reasons for people to get excited about their next phone or device. Point being, it's not as apples to apples as it might seem. For now, though, Nvidia does not seem like it's slowing down anytime soon, although, to be clear, none of this is financial advice. Next up in the politics of AI, Treasury Secretary Janet Yellen is warning that AI poses significant risks to the financial system. The candidate of the speech was the very traditional, lots of opportunities but also risks, kind of a thing. As CNN business writes, on the opportunity side, Yellen will note how AI has already been used by investors to support forecasting and portfolio

management and by banks to fight fraud and support customer service. However, on the risk side, Yellen is worried about complexity and opacity. She's worried that AI models operate as a black box, and that, quote, if Wall Street firms are relying on mysterious AI models, regulators will struggle to understand how safe their systems truly are. Yellen also plans to cite, quote, inadequate risk management frameworks around AI risks and interconnections that emerge as many market participants rely on the same data and models. This is something that SEC Chair Gary Gensler has been talking about as well, basically that if everyone is relying on the same models to tell them what to do next, that could exacerbate market moves both on the way up and on the way down. I've spoken about this

before, but I think this concern is remarkably overblown. First of all, there's no universe in which everyone is just going to use the same off-the-shelf model. And second of all, Wall Street's entire job is to find unique pieces of data and approaches that give them alpha. I anticipate they being some of the leaders in customization of models that come up with different answers. And to me, it feels like this concentration risk is very much in the realm of the theoretical. However, if you take nothing else away, it's clear this is on the radar for the biggest financial regulators in the world. Over in the world of AI devices, humane continues. to have trouble. The latest is that Humane has warned AI pin owners to, quote, immediately stop

using the charging case that came with the pin because issues with the third-party battery cell, quote, may pose a fire safety risk. The company is offering two free months of the subscription as a sorry. Finally, today, an interesting new product that lots of people are chatting about is Asana's new AI teammate. Asana is, of course, a team task management platform, which helps coordinate around different responsibilities. And the idea of these new AI teammates is that they'll be able to actually use information about historical relationships and past projects to go ahead and assign work to people who have the best matched skill sets. Said Asana co-founder and CEO, Dustin Moscovitz, were able to do this better than anyone else because we built Asana on the work

graph, which maps the relationship between the work your team does, the information about that work and the people doing the work. By way of examples, Asana pointed to one unnamed marketing organization where, quote, it says the AI bot is putting together tailored marketing content, translating assets into different languages and standardizing workflows. The AI teammate will come with a chat-based interface so users can ask it questions, and all of this amounts to yet another push into the agenic space, where these tools are just going to get more and more sophisticated and better able to start delivering on more complex tasks. For now, though, that is going to do it for the AI Daily Brief Headlines edition. Next up, the main episode. Today's episode is brought to you by

Super Intelligent. Regular listeners know that Super is our platform for helping people learn how to actually use AI tools. These are not long, laborious courses. These are fun, fast tutorials that get you actually using the world's most interesting and useful AI tools within minutes. If you want to build a web application with no code, we've got tutorials for that. If you want your presentations to look better than ever and take you less time than ever, we've got tutorials for that. If you want help brainstorming, writing social media copy, and just generally working smarter, faster, and better, we've got tutorials for that. We've worked really really. hard to make it so that there is no better place on the internet to learn how to actually put

AI to work for you, and I'd love for you to check it out. Go to be super.a.i and use code podcast for 50% off your first month. Once again, that's besuper.a.i. Welcome back to the AI Daily Brief. One of the realities of artificial intelligence so far, generative AI specifically, is that it seems to have concentrated more power in the hands of big tech. We've talked frequently on this show about how the high cost of compute has meant that a very small handful of companies can actually do so. How even Silicon Valley venture capital isn't necessarily enough capital for companies to compete at the state of the art. It's why we've seen these massive deals like Microsoft's 10-figure investment into OpenAI, Google and Amazon both plowing billions into Anthropic. There are other dimensions,

though, that are also concentrating power. The fact that big enterprises that are plugging in their data to AI systems want to work with partners that they already trust creates even more momentum for the incumbents. And there are lots of people worried about this concentration of power. One group are the open source advocates who fear that overzealous regulation too early will lead to regulatory capture for the players at the top of the heap. But the U.S. government is also concerned about this concentration of power. And after some wrangling, the Justice Department and the Federal Trade Commission

have now figured out and agreed how to divide responsibility for investigating three different players in the AI space for antitrust violations. Basically, the Justice Department and the FTC were both jockeying for positioning as related to investigating these players when it comes to potential antitrust issues. That means that they were delayed in having to sort out who was going to have jurisdiction over what. And according to reporting from the New York Times, the Justice Department will take the lead in investigating Nvidia, while the FTC will focus on OpenAI and Microsoft. This is not the FTC's first foray into the space. In July, they opened an investigation

on whether OpenAI had harmed customers through its data collection practices. Back in January, they started a broad inquiry into these strategic partnerships between AI startups in big tech. Invida is perhaps new in some ways, but it's not necessarily surprising. Writes the Times, industry players have grown worried about Nvidia's dominance, two people with knowledge of the concern said, including how the company's software locks customers into using its chips, as well as how Nvidia distributes those chips to customers. The Justice Department has also been signaling their interest in this area. The Times piece concludes, Last week, the Justice Department's antitrust division organized a conference about AI at Stanford

University. In his opening remarks, Jonathan Cantor, the top antitrust official at the agency, pointed to, quote, structures and trends in AI that should give us pause. He continued, AI relies on massive amounts of data and computing power, which can give already dominant firms a substantial advantage. Cantor continued those comments today in the Financial Times as well. He said in an interview with F.T, that they are examining, quote, monopoly choke points in the competitive landscape and are concerned that the nascent AI sector is, quote, at the high watermark of competition not the floor. The interview gives a little bit more, on how they might think about some of these issues. From the FT, Cantor pointed to government initiatives

to boost domestic production, including the $39 billion of incentives in the Chips Act, but added that antitrust regulators were looking at how chipmakers decide to allocate their most advanced products amid rampant demand. Said Cantor, one of the things to think through is conflict of interest, a thumb on the scale, because they fear enabling a competitor or helping prop up a customer. If decisions are being made that show companies are not caring about maximizing profitability or generating shareholder value, but more looking at the competitive consequences, then that would be an issue. Another big question seems to be the sort of aqua hire that happened with Microsoft and Inflection. When most of the 70-person team at inflection left to join Microsoft,

which then turned around and did a $650 million licensing deal with the company that effectively bought out their early investors, many people thought it was Microsoft trying to get around potential antitrust concerns. Well, that clearly didn't work. As Cantor said, aqua hires are something that antitrust enforcers will look at. We're not using stylistic or formalistic characteristics of how these companies explain these deals. We are looking at the market realities. We are focused on the facts. If the form is different but the substance is the same, then we will not hesitate to act. We look at what are the raw materials to produce a product, whether that's steel or engineers that fits within the traditional paradigm of what we care about. Microsoft for its part

pushed back on this, with President Brad Smith telling the FT, the partnerships that we're pursuing have demonstrably added competition to the marketplace. I might argue that Microsoft's partnership with OpenAI has created this new AI market and without its help, OpenAI would not have been able to train or deploy its models. Adam Kovacevich on Twitter gave a little bit more background on this horse trading between the Department of Justice and the FTC. He said at the past as a guide, the deal is likely to result in AI-related antitrust lawsuits as early as next year against Microsoft, invidia, and OpenAI. Adam continues, Most people know that the FTC and the DOJ share responsibility for antitrust enforcement.

There used to be a quaint time when the agency split antitrust investigations and cases along industry lines. This approach had two virtues. Agency staff could develop expertise in industry areas, and investigation started by looking at competitive dynamics within industries, rather than, say, picking a prominent company the agency wanted to sue and then figuring out a case. This all broke down over big tech. Both agencies had brought cases in tech, but it was a big, juicy headline-grabbing target, so FTC spent the Obama and Trump here sparring over turf. This resulted in a 2019 handshake deal. FTC would take Amazon and meta, DOJ would take Apple and Google. This wasn't a principled way to approach antitrust enforcement. It was a target-driven approach. The result was unsurprising. The agency's

weren't going to spend all that time sparring, reach a deal, and then not bring antitrust cases. That would look weak. A year later, FTC and DOJ brought their first cases under this deal. FTC versus Facebook and DOJ versus Google. So far, the government's record on the cases brought under this deal. Zero wins. One loss with meta. One motion to dismiss lost.

Meta, Instagram, WhatsApp. One trial in progress, Google search. Four cases not yet at trial stage. A problem facing all these cases, the agency started with the target companies, decided they wanted to bring cases and then figured out the best cases to bring. Rather than looking at industries overall, they found cases to fit their targets. This has resulted in the agencies contorting themselves to define markets and judges rejecting and being skeptical about those definitions.

Flash forward to 2023 in the rise of AI. For the last year, FTC's Lena Khan and DOJ's Cantor have given speeches about the danger of AI monopolization, despite the A, nascent quality of the industry, and B, the incredible competition among both incumbents and upstarts. These speeches were really about turf, trying to position their respective agencies to lead on A. Now, Khan and Cantor have reached a deal where DOJ will add Nvidia and FTC will take on Microsoft and Open AI. Is this a principled way to approach antitrust enforcement in AI? No, it is a target-based headline-seeking approach. Based on the previous deal, it will almost certainly result in

government lawsuits. But so far, DOJ and FTC's myopic approach has been there undoing. Robert Sterling summed up the feeling of a lot of people when he said, the federal government wants to kill the AI industry in its infancy. By launching antitrust investigations at such an early stage, they're sending a signal that they intend to regulate the industry with a heavy hand. Ironically, this will reduce the flow of capital into AI, resulting in fewer startups, reduced competition, less disruption, and even more dominance for today's incumbents. But I wouldn't expect career DOJ or FTC bureaucrats to understand this. Max Meyer from 8VC points out how this has worked in the past.

He writes an antitrust story in three parts. One, Adobe to buy Figma for $20 billion, amazing for customers, prices low forever. Two, DOJ, no, monopoly. Three, Figma doesn't make enough money to IPO, has to raise prices. is significantly. Well done, antitrust bureaucrat morons. Even more on the point is Darrow Boshenjo, who writes, investigating invidia or open AI for being too dominant in AI makes it seem like these antitrust investigations are about punishing success. Breaking up Nvidia or open AI sounds absurd, especially when the industry is fairly nascent. As you can probably tell, my biases lie with

these folks who think that this is not a particularly good use of government resources at this time. It's not that I don't think concentration of power is an issue. It's more that I don't think the remedy is government regulators, and particularly antitrust regulators at this stage. Capitalism is doing its job right now. If you look at the market for chips, yes, Invidia is dominant, but they have never faced more of an all-out assault from so many different angles. And when it comes to LLMs in the underlying models, the single best thing that governments could do

would be to support open source efforts, but that seems like exactly the opposite of what they're going to go for. Ultimately, this is just the beginning of jockeying for power when it comes to AI in government, and I think it's going to get a lot messier from here. For now, though, that is going to do it for today's AI Daily Brief. Until next time, peace.