Does xAI's $6B Raise for Grok Make Sense?

0:00Today on the AI Breakdown, is the$6 billion XAI investment deal crazy? Before that on the brief, how Wall Street is thinking about the value of AI right now. The AI Breakdown is a daily podcast and video about the most important news and discussions in AI. Go to breakdown.network for more information about our YouTube, our newsletter, and our Discord.

0:24Welcome back to the AI Breakdown Brief, all the AI headline news you need in around five minutes. And we have a very Wall Street slash big tech centric AI Breakdown Brief today. We have one big tech stock that has cratered on AI plans, one tech stock that is soaring around AI, and one big tech stock that's doing pretty well, although how much it's AI versus other things isn't exactly clear. Let's start with a stock that's doing well, which is Microsoft. In reporting their last quarter earnings, Microsoft beat Wall Street estimates. Analysts had been projecting$60.80 billion of revenue and earnings per share of$2.82, cents, reaching$61.9 billion in revenue and$2.94 earnings per share.

1:03That is a 17 % jump in revenue. All told, the company's shares increased by more than 4%, increasing the company's market value by$128 billion. Now, right at the center of the story of why they were doing so well was, of course, Genitive AI. Said Brett Iverson, Microsoft's vice president of investor relations, we're continuing to see customer demand grow quite a bit, and so we're making sure to scale our available capacity in line with that. Now, of course, where this comes home to roost is in Microsoft's cloud business. That's the division through which they're offering these AI capabilities.

1:33And if you look across the last year, you see that Microsoft's Azure has flipped and grown ahead of Google Cloud and is also significantly outpacing AWS. Revenue from Microsoft's cloud unit rose to $26.7 billion, which once again passed Wall Street analyst bets. Azure revenue overall grew 31%. And Microsoft forecast that its growth for the next quarter would be 30 to 31 percent with revenue of$28.4 to$28.7 billion, which would put it ahead of the 28.5 percent Wall Street growth target. One of the things that has been really notable about AI as a new technology is that because it is being used inside of these cloud businesses as a way to attract buyers to these cloud services, it actually has an outlet that Wall Street can understand in dollars and cents terms right now.

2:16In other words, Wall Street doesn't have to be super forward-looking. It can understand the value of investing in AI right here in this moment. And indeed, what we saw around Meta stock suggests once again that Wall Street is not the same as venture capital. They are not imagining what might be in 5 or 10 years, but instead they are looking very much at the here and now. Meta also reported last quarter's numbers. First quarter profit more than doubled year over year, and revenue was up 27%. So why then did Meta's stock fall by as much as 15%, wiping somewhere between$180 and$200 billion off its market cap?

2:49Well, it seems like it was all about its projected AI investments. Meta increased its predictions for what it would spend on AI investments by$5 billion this year. Previous guidance had suggested that Meta's full-year capital expenditures would be in the range of$30 to$37 billion, and now that's moved up to$35 to$40 billion. Said one analyst, the language around spending plans has become bolder once more, and this could be what's spooking markets. For all Meta's bold AI plans, it can't afford to take its eye off the nucleus of the business, its core advertising activities. Meta's resources are vast but not infinite, and its digital advertising market share needs defending at all costs.

3:23Now, it wasn't just that Meta was talking about its big investment in AI. Zuckerberg was also preparing investors for the reality that he saw, saying that they would be increasing spending quote meaningfully before we make much revenue from some of these products. He did also note that the company has a quote strong track record of monetizing new services effectively. But ultimately, as we've seen over and over again, Zuckerberg is going to do what Zuckerberg is going to do. Wall Street and everyone else be damned. Then there's Google. Google is up nearly 11 % as I record this, after reporting a 15 % rise in revenue from a year before, which was more than 13.5 % rise the previous quarter.

3:58Now for this one, the increase in the stock price seemed not to be really about AI. Instead, Google made a dividend and a buyback announcement, and that seemed to be more of what was encouraging this jump. Indeed, as the Wall Street Journal writes, investors are closely watching for signs of whether and how the tech giants are reaping new business from investments in AI programs that can generate text and images. Google, in particular, has faced persistent questions about its ability to defend its business model, which relies heavily on search ads as more people turn to chatbots for answers. Google has offered few specifics about how AI is affecting its sales and profits.

4:29Multiple analysts asked CEO Sundar Pichai on a call Thursday for more details on how AI is improving Google's financials. Pichai said Thursday that the company is encouraged by an increase in search usage among people trying its new AI tools, but he otherwise provided little in the way of metrics on the technology's business impact. The role that these big tech companies are playing in an emerging technology movement like AI is really different than what we've seen before. The normal pattern is that some new types of startups create some new types of usage patterns, which eventually get acquired into big companies or become the big companies themselves, but the sheer capital needs in the AI space is making this a little bit different.

5:06There are other factors beyond capital as well. For example, I think large enterprises, because so much of the value of AI lives inside how it interacts with their data, are more inclined to trust companies that they're already working with rather than new startups. Still, part of the reason that I spend so much time on the ebb and flow and the fortunes of these companies is the way that they treat AI will have a big impact on how the rest of the world does. For now though, that is going to do it for today's AI Breakdown Brief. Next up, the main AI breakdown. Attention, AI Breakdown listeners.

5:35Consensus 2024 marks the 10th gathering for all things crypto, blockchain, and Web3. However, importantly, this year's agenda will also dive deep into AI-driven transformation. And the speaker lineup includes the leading minds and innovators at the forefront of this digital renaissance. Don't miss the Consensus AI Summit to cut through the hype to find where true transformation and opportunity lie. Listeners to this show can get 15 % off registration with the code AIBreakdown. Visit consensus2024.coindesk.com to learn more. Some of the folks who will be at Consensus this year include Guillaume Verdun, aka Beth Jezos, founder and CEO of X-Tropic, as well as spiritual leader of the accelerationist movement, Neil Stephenson, co-founder of Lamina One, and Brendan Eich, the CEO of Brave Software.

6:16Again, go to consensus2024.coindesk.com to learn more and get 15 % off registration with the code AIBreakdown. Before we get back to the AI Breakdown, I want to share something fun we have coming up on Superintelligent next month. Superintelligent is, of course, our new platform for teaching people how to use AI in a way that that is much more fun, fast, and practical. The platform has hundreds of short tutorial videos, each of which is paired with a set of step-by-step instructions that get you using AI tools in minutes, not hours, and certainly not days. For those of you who haven't signed up yet but want to check it out, in May, I am running a special NLW cohort.

6:51What this means is that people who sign up with the code NLWMay will get$5 off their first month, but they'll also have access to a private channel in our Discord with me. I'll be handpicking tutorials each week that I think are the most useful to start with. And I'll also be available for questions, advice, and feedback from this group. Spots for this cohort are limited, so if you want to be a part of it, again, sign up at besuper.ai with code NLWMay. That's besuper.ai with code NLWMay. Welcome back to the AI Breakdown. We kick off today with the news that Elon Musk's XAI is close to closing on$6 billion.

7:29Now, Elon, for his part, has been very circumspect about this fundraise. Rumors of it started months ago, but he denied them, said that they were not actively raising. However, the reality of the situation has been fairly open, and for a long time it appeared that the number was around$3 billion. Over the last couple weeks, that seems to have doubled to a$6 billion raise on an$18 billion pre-money valuation, meaning that after the raise, the company would be valued at$24 billion. The information is putting a little more meat on this bone, citing two sources who are close to the deal. The additional information that they give us include some of the folks who are investing.

8:04Sequoia Capital is one of the investors participating in this round, which is notable because, as we've seen over and over again, a lot of the capital that has flown into the foundation model space has not been from traditional VC, but from big tech companies. The reason being, of course, that traditional venture funds don't have the ability to do 2, 3, 4, 5, 6 billion dollar rounds. However, for Elon and XAI, they obviously can't turn to a Microsoft or a Google when they're trying to beat Microsoft and Google. So it'll be interesting to see what exactly the full composition of this round ends up being.

8:33As the information points out, if this really does come together the way that it looks like it will, it will represent one of the largest single funding rounds among these foundation models, probably just behind the OpenAI Microsoft investment. Certainly, Elon Musk has not been cagey about the fact that they need more resources to compete. So far, Grok has released 1.5, and Elon has said that Grok 2 is coming, which is currently being trained on 20 ,000 NVIDIA H100s. However, earlier this month on a Twitter Spaces, Elon suggested that the company needs 100 ,000 GPUs to train Grok 3. So for what it's worth, this is a deal that I've been asked my opinion on a number of different times, especially from investors who are outside of the AI space.

9:11And so what I want to do today is kind of go through the logic that I've given, although to be clear, there's no financial advice here. It's just a way of looking at this. The TLDR on what I've said is that there are a very, very small number of companies that are legitimate competitors for the generalized foundation model space. And if that is a space that you want investment exposure to during these private days, you kind of got to pay what it's going to cost to play. So who are those companies? Well, of course, there's OpenAI. At this point, OpenAI is valued in the$100 billion range. And the most recent rounds have only been tender offers of employee shares, limiting how much is actually available.

9:47Point being that while it's not impossible that there will be future raises that investors can participate in, OpenAI is not the easiest company to get equity in right now. And even if you do, it's at a fairly significant premium. Anthropic has been raising left and right. In 2023, they raised$7.3 billion, which includes the$4 billion Amazon commitment, which was actually just finished earlier this year in 2024. One of the interesting things about that story was just how these astronomical sums have an impact of crowding out smaller investors. And by smaller investors, I'm not talking in normal Silicon Valley terms.

10:20Back in February, the Times wrote, even after raising billions from Amazon and Google, Anthropic knew it would eventually need more money. Generative AI startups are constantly updating, refining, and expanding their technology to make their product accurate, up-to-date, and more powerful. And that requires enormous amounts of expensive computational power. Finding new investors was easy for Anthropik, but many of those who were interested wanted to invest$10 million to$25 million, while the company aimed for a much larger sum. Basically, if you're trying to raise a billion dollars, do you want one or two companies who can do the entire amount, or do you want 100 investors putting in$10 million each?

10:52From a sheer logistical and administrative perspective, obviously the fewer investors, the better. Now, when push came to shove, the way to accomplish what Anthropic was looking for, while also allowing smaller investors in, was to roll up all those smaller investors into a special purpose vehicle that was led by a single firm, in this case, Menlo Ventures. Anyway, again, the point of this for our story is that barging your way into that deal was also not trivial. Beyond OpenAI and Anthropic, who are the other realistic contenders in this space? Well, of course, there is Google with Gemini, but for exposure to that business, you just have to buy public stock at this point.

11:24Same goes with Meta and Llama. And Meta and Llama are throwing an entirely different wrench into this, which I'll come to in just a moment. One more credible competitor, which honestly kind of looks like a steal given everything that we've talked about so far, is France's Mistral, which has exploded onto the scene over the last nine months or so. In December, the company raised at a$2 billion valuation, and earlier this month, it was reported that they were seeking funding at a$5 billion valuation. So that brings us back to Musk and OpenAI. So one part of the logic to participate in this deal is simply the fact that many people are looking for exposure into this space, and it's one of a very small handful of options.

11:59I will note at this point that there are a number of other really interesting global competitors in places like China that are not primarily English-based LLMs, but obviously from a U.S. investor standpoint, which is going to inherently be the bias of this show given that I'm sitting in New York, these represent a fairly big portion of the going options in many ways. Now, let's come back and talk about Lama 3 for a minute, because I think that the emergence of Lama 3 actually works in Musk's favor when it comes to demand for and the price of XAI. One of the big things that we've talked about on this show over the last couple weeks is the extent to which how close Lama 3 is getting to GPT-4 is changing the way that developers think about access to models.

12:36Basically, I think that people are starting to wonder, will state-of-the-art models just simply become a commodity? Or perhaps more acutely, will Zuckerberg be willing to spend enough to force them to become a commodity by giving away Llama 4, Llama 5, etc.? The answer increasingly appears to be maybe, which I think creates a real challenge for the anthropics and open AIs of the world who potentially are going to have a harder and harder time with their enterprise models when they're competing with lower-cost open-source options. Now, that's not to count them out at all. There's a million other dynamics which could still matter, and there are futures in state-of-the-art which could still create massive differentiation, which continues to command a premium on price.

13:11But in a world where there is more commoditization, let's say, of highly advanced models, it sort of makes companies that have an existing distribution platform, a business model that these AIs plug into, look even better by comparison. An example of this is the new meta AI that sits at the top of Instagram, WhatsApp, and Messenger. When Llama 3 came out, they didn't just release it as a standalone chatbot. They embedded it directly in their three very popular messaging apps in a way that totally synced it up with Search. It's early days, but I think dramatically more people are going to use it based on those product decisions and actually have a positive experience with it.

13:47Which brings us back to XAI. So if reason one to participate in this type of round is simply the lack of availability and the scarcity of rounds like this, another reason might be a bet on the integration with Twitter slash X. Even if Grok isn't state-of-the-art in anything else, the fact that it is embedded in this larger network, that despite all the hemming and hawing about Musk's leadership is still one of the very small handful of very important social networks in the US and around the world that creates more of an opportunity for Grok to add value to consumers and potentially plug into existing business models than perhaps the standalones like OpenAI and Anthropic.

14:23Now, of course, we don't know how any of this is going to play out. We don't know how much the distance between state-of-the-art and open-source alternatives will matter. My point is simply that as big and eye-popping as these numbers seem, they're a lot less crazy than you might think once you really dig in. For now, though, that is going to do it for today's AI Breakdown. Until next time, peace.