Limitless Podcast - THIS WEEK IN AI: Massive Quarterly Earnings, Anthropic's Project Deal, Cursor SDK

Episode Date: May 1, 2026

We tackle the strong earnings reports from Google, Amazon, Meta, and Microsoft, propelled by AI and cloud growth. Google sees a 94% earnings per share surge, while Amazon's AWS boasts 28% growth. Meta raises revenue by 33% through AI advertising, and Microsoft reports a 40% increase in Azure. We also discuss innovative AI projects like Anthropic’s transaction trial and 11 Labs’ music licensing.------🌌 LIMITLESS HQ ⬇️NEWSLETTER:    https://limitlessft.substack.com/FOLLOW ON X:   https://x.com/LimitlessFTSPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQAPPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890RSS FEED:           https://limitlessft.substack.com/------TIMESTAMPS0:00 Earnings in Tech History3:41 Google6:22 Amazon10:15 Cash Flow14:47 Meta18:16 Microsoft22:22 Apple26:07 Anthropic Project Deal28:06 Cursor's AI Agent Harness30:06 Eleven Labs38:16 Closing------RESOURCESJosh: https://x.com/JoshKaleEjaaz: https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠

Yesterday, four of the five most valuable companies in the world reported the biggest earnings in tech history within a 19-minute window. And now we have a pretty good idea of whether we're in an AI bubble or not. Google, Amazon, Meta, and Microsoft reported their Q1 quarterly earnings, and it blew every single Wall Street analyst's expectations out of the water. Seriously, these numbers are pretty insane. Google's earning per share was rated 94% above expectations. There's not a typo.

94% and every single cloud business, AWS and Google Cloud reported revenue increase average of 50% with Google having back orders of half a trillion dollars worth of compute orders. Even Mehta, who's categorically been behind in the AI race, reported massive growth in revenue for their ad business thanks to AI. In fact, it's so good that they're beating Google at their own business of search advertising. For two years, AI critics have constantly been saying that we are in an AI bubble. They're saying the House of Cards is going to come down that we're spending too much. The earnings reports of yesterday proved the exact opposite. All these companies that are spending hundreds of billions of dollars are actually making more

money than they ever did before. So are we in a bubble? The answer to that question is probably unequivocally yes. I think we are to some extent. But what is the size of that bubble? Well, the chances are it's much smaller than I think a lot of people think. And how much runway is left, well, I think a lot more than people anticipate. And we could start with Google, because that's what you're mentioning a lot. I mean, their earnings report showed, we got to start with the CAPEX number. So for those who aren't familiar with CAPEX, KAPX is the capital expenditure, the amount of money that a company is going to spend over the next X period of time. Now, for Google, they provided CAPX for their next calendar year, and they guided even higher than they did previously

at $180 to $190 billion. This is a tremendous amount of money. like a stupid, crazy amount of money. Their revenue, for reference, $110 billion is more than Coca-Cola's annual revenue. And this is what they did in a single quarter. One of the most kind of worried about and concerned about categories, Google Search, I mean, I remember a few months ago very vividly. We were on the show talking about Google Search and whether or not it's going to be a viable business. Is the chatbot LLM interface going to cannibalize Google Search? The answer now is an unequivocal? No. Google Search grew 20% to 60 billion. billion dollars, which is, again, just a huge amount of money. And I think the headline here is that

capital expenditure is going up, meaning the money is going to be spent on building out the infrastructure required to scale AI at the same time that they're actually generating a lot more profits from their services like Google Search and Google Cloud. I think one of the themes we're going to notice as we cover these earnings reports is that cloud revenue is kind of the bread and butter of a lot of these companies. And it's accelerating very quick. I mean, for their projected guidance on their cloud revenue that they're going to make next year, they doubled it to about $460 billion. And for reference, that's larger than the GDP of Belgium and roughly equivalent to two of Walmart's annual revenues just in cloud search alone. So the Google numbers are

incredible. The one thing that really sent me for a loop in terms of how I kind of calculate the guidance of where we are in the say I bubble is a line that Sundar said in the earnings report, CEO of Google, who said their 2027 KAPX will significantly increase in 202027 next year, compared to this, which is already a record. So they are really guiding towards a very strong 2026, a strong 2027. And that was one of four that really made a pretty big impact on the market. Well, I just want to tell a slightly different story on the Google side, because really it's that cloud business that absolutely killed it. They made $20 billion in the market. And the last three months for guidance here that's around 63% more than they did this same time last year.

So it's a massive jump up, but then you might be wondering, well, how amazing is this? How substantial is this? The projections that you just mentioned, the half a trillion dollars, so it's actually $462 billion, I believe, is already locked in orders. It's not a projection. These are orders that they already have on the backlog, and people are waiting for this to happen. And in fact, I have a neat little graphic, which shows this, because this caption goes, this is so crazy. It literally looks fake.

And it shows you the back order. $460 billion just for Google Cloud alone. This is one of like eight different major business units that feed into the Google profit funnel. So I just want to emphasize that. Now, Sundar also said on the call, we, and I'm quoting this, we are compute constrained in the near term. Our cloud revenue would have actually been higher if we were able to meet the demand. So the point is, they don't have enough supply TPUs and infrastructure, which is, by the way, Google Cloud is one of the biggest cloud providers in the entire world. I think they're like second to AWS, Amazon, which is the biggest one to talk about in a second.

They can't supply enough. Google is like one of the biggest hyperscalators that are out there and even they can't meet demand. So the point is the criticism that we're faced around AI bubbles is we're investing in all this AI cabex. Google alone is investing, I think, $150 to $200 billion this year to create more. TPUs and AIM for it to supply the demand. The skepticism has been that there is no demand for this. No one's willing to pay for this. These quarterly earnings prove that categorically incorrect. And we have a half a trillion dollar backlog. Yeah, there's no way that the spending is going to slow down. And the fact that the only constraint that they had was compute, not demand, given the profit margins that they already

have is pretty telling. I found one funny fact about this is that a lot of the revenue actually came from their private investments getting marked up. So of their total net income of $62 billion, 37 of it was unrealized gains on equity securities that includes companies like SpaceX and Anthropic. You mentioned the largest cloud provider, Amazon, who also reported yesterday, they have a similar thing where I think of their net income of $30 billion, $16.8 of it was their pre-tax gains on anthropic shares. So this is very much a circular economy where they're investing and Anthropic, the anthropic valuation is rising, a large percent of the net income becomes anthropic, becomes SpaceX. So earning season when those IPO is going to be pretty crazy,

but we have to talk about Amazon. Their AWS services is up 28%, which is their fastest growth, they said, in 15 quarters. That's a $150 billion run rate. To put that in context, that's larger than the entire annual revenue of Disney, Netflix, and Spotify combined. And that's AWS alone. It's bigger than the entire global movie industry at a margin of almost 40%. So again, this is the same story we heard with Google. Their cloud is really just constrained by compute. Their margins are high. They have more demand than they ever have before.

And now we're seeing them move into building their own chips and working with other companies. And now that Open AI is not locked in with Microsoft, we're seeing them start to serve open AI infrastructure as well through their bedrock. It's a really impressive range report from Amazon as well. There's been a trend shift with Amazon in particular. Now, if you talk to an average person about Amazon a few years ago, they'd be like, oh, yeah, it's that e-commerce company. It's where I order my toilet paper from. Amazon has undergone a very categorical shift into the compute infrastructure and distribution service. Now, that sounds like a lot of buzz terms. Let me explain what that means. AWS is basically the bedrock for any internet company that exists today. If it runs online, it's most likely running off of Amazon's AWS servers. And so what that means is basically,

is, okay, you can run a website, you can service all the requests that you have from people trying to buy stuff from you. But with this recent AI shift, Amazon looked at it and thought, huh, this looks really similar. We could be the distributor and infrastructure providers for AI. And guess what they already had? A bedrock of customers that already had that. And I use bedrock very specifically, by the way, because their main AI part is literally pun, very much intended, was called or is called Amazon Bedrock. And just over the last week, they now officially are going to be the main distributor for OpenAI's chat GPT models. The reason why that's a big thing is because Microsoft was the sole cloud provider for Open AI.

But this week, that contract broke, which meant that Open AI can now serve their models through whatever distributor they want. And the first door that they were knocking on was Andy Jassy at Amazon saying, hey, we would love to be the main provider. Why? Because Amazon has connections to all the enterprise companies. They service all the government's AI requests and all the Fortune 500 companies, AI request. They run all the servers. No one wants to pick the model. No one wants to set up the dev environment. AWS has got you and they are creating a crazy business for this. Like you mentioned,

28% growth. That's almost $40 billion they've made over the last three months. The fastest growth in the last four years. That is not an anomaly because in Q4, we were saying the same thing when we were reporting on their earnings as well. And it's good to see that the market is now finally starting to understand that. And they this week actually breached their all-time highs. I think it breached over 260 price per share. which is insane. And the final point I want to make about Amazon and Google, actually, is there's a common theme.

Now, if you look at total revenue earned by both of these companies, it's increased by a decent amount, but not like mind-blowing amounts. Where you need to look is the profit margins that both of these companies are made over the last three months. That has expanded, on average,

by around 50%. I want to re-emphasize that point. 50% for their cloud businesses, which means that AI chips, which is what both Google and Amazon provide and their cloud services, allows them to eat more profit on the margins, which is very good if you're any kind of a shareholder of either of these companies. Yeah, and the pricing, UlaS. Sissy is very high too. It's not like they are out of customers. They just don't have the compute.

So theoretically, they could raise these prices, increase those margins quite a bit, and people will still continue to pay. I know for a fact, like someone like OpenAI can double the cost to serve it, and Amazon could forward that on to the customer, people would still pay. And Amazon's CAPEX, again, at $200 billion. To put that into perspective, it's larger than the entire annual CAPEX of the U.S. oil and gas industry. Like, this is a huge amount of money. And then we have one other quote from the earnings report, which I thought was fun. It was Andy Jassy talking about the combination of their industry leading price performance, Traneum, custom silicon, market leading gravitons, CPUs, and their core AWS. It, he, he's basically describing as a once in a lifetime opportunity. He said, Amazon and his shareholders

will be very well off in the future as Amazon crossed a new all-time high. So it's very exciting to see all of these numbers go up. And we don't even, we didn't even talk about their logistics business. Like the part of Amazon that we interact with on a daily basis when you order groceries or order things from Amazon, it's just an unbelievable business. There's like 10 different businesses. And I think that's the theme of a lot of these larger mega cap companies is as they take, as they have the else cash flow coming in, they're going to spend it efficiently and they're going to increase their margins elsewhere. What an incredible business. Well, let's speak about the cash flow for a second because we sound very rosy right now and that it's like the best companies in

the world. Now, I want to point out two things. As of the end of Q1, Amazon has now spent around or committed to spending 94% of their free cash flow. We're talking about like money that they have available coming in through operating income and through money that they have on the cash balance sheet. out of the four companies that we're speaking about today, they are the most aggressive spenders on AI infra. And the bet that Andy Jassy, the CEO was making was we're doing this intentionally because we know for a fact that we're going to make money from this. And the projections or the reported earnings from Q1 has proven that exact point. But it's still a very aggressive maneuver, right? Like as of the next three months, you could argue that either if they

don't have enough money or they run into some kind of hurdle, they might need to start having debt financing and start borrowing money. Now, as of today that we're speaking about it and reported on these earnings, they aren't in a levered position. So for everyone saying that we're in an AI bubble, an AI bubble by definition means that these companies are levered up on their balance sheet. Neither of the companies that we're talking about today, Google, Amazon, and Meta and Microsoft which we're about to, are levered. So like I disagree with your earlier point. Josh was basically, I don't think we're in a bubble yet. Big emphasis on yet. Well, I think we're definitely aligned on being bullish about where we are

in this stage of the market, whatever it is, that there's a lot of runway left. But there's an interesting dynamic to that point around free cash flow, which is the money that basically is left over after all their expenses. A lot of that is flowing out. Like you mentioned, about 95% of Amazon's earnings, the free cash is flowing out. And this is kind of inverting what we've traditionally had in the markets. Generally, Amazon, Google, these companies have absorbed all of the cash and they have a tremendous amount of free cash flow, now they're spending it. You're seeing a lot of money going out the door. The CAPEX is between the four companies we're going to cover is about $600,700 billion. The question is like, where is that money flowing to? Because the reality is that while

cash flow is coming out of Amazon, it is finding its way into other companies as well. And we're starting to see this with companies like Bloom Energy, who we famously mentioned on the show a few months ago and then a few weeks ago and still have not filled large enough bags to see the upside of this, which is very frustrating, but Blue Energy is up 1,400% on the year. And this is one example of where that free cash flow is going. When companies are building, they're spending this CAPEX, they need to create data centers, they need the shell, they need the energy, they need the chips. Bloom Energy is an energy part of that equation. And it's clearly captured a lot because it was just up 24% yesterday on huge earnings. And now Leopold, in the most recent 13f filing, he had a position of $875 million,

is now in a position worth $2.7 billion. This is one example of the money. The other, I guess we have Sandisk that we should probably mention briefly. If you invested in Sandisk two years ago, you're 30 times your money. Like 30x. Every thousand dollars you put in is $30,000. Why is that? Well, because one of the critical thresholds and something that we heard on all of the four earnings reports that we're going to mention is that memory is a serious supply constraint. And the costs associated with that are going up. And the companies are just absorbing that. I think it was maybe Amazon or meta who added $25 billion in CapEx just for the increased costs. of these goods. And companies like Sandisk are the ones that are absorbing those. So we're seeing

for the first time in a while the money flowing downhill from these like mega cap companies down into the core infrastructure. And I think that's an inversion that hasn't quite been priced in because it hasn't been that way in a really long time. Exactly. And so the theme that we're spoken about for the companies so far, Google and Amazon, has been very much on AI CAPEX spending, on Google Cloud, on Amazon web services and whether there is customer demand for the infra. Now, another company that we want to speak about today is meta. And in fact, they're the company that added an additional $35 or $25 billion for CAPEX to their KAPEX spend this year.

They're playing in a bit of a different field. Like, meta is, for better or worse, a social media company, but they're really, really good at making money from advertising. And for the longest period of time, it's been Meta versus Google. And Google has, quite frankly, always won. They've dominated, they're the doormat to the internet, right? They have dominated search. and meta has come close but never quite as close, right?

Except for the last three months where they started integrating AI, specifically their new models and Manas who they acquired for $2 billion, which just got blocked by China. That's a whole other thing. You should go check out our episode that we published earlier this week. The point being is meta started integrating AI, and they started having higher AI search conversions. Now, I need to make it very clear.

SEO, search engine optimization, is a huge business. And companies and advertisers pay billions and billions of dollars per year just to get access and put their product in front of the right profile of person. Meta is pioneering the AI version of that. And we were kind of in the dark. They were kind of behind on the AI race. They don't have a frontier model. So we didn't know whether they could actually pull this off.

And the data has shown that they can. They beat earnings per share by 53%. And their revenue is up 33% year over year. which is just categorically very impressive. And I think we're going to start to see more of these things happen with other such companies. Google being the most obvious one where everyone thought AI was going to eat into their search revenue and they're up 20% over the quarter.

So overall, I think meta, it's not just an AI CAPEX infra thing. It's also AI bleeding into every single other type of product category and also making them money. So the demand for this thing for this technology is insatiable. And the lesson that I'm hearing from all these three companies that we've spoken about so far is there's not enough compute to go around. Okay, so Ejos, I'm hearing you say all these positive things about meta, and then I'm looking at the stock today and is down 9%.

So clearly there's a disconnect between the market and the numbers. And we've seen this before. People sell on the news, even when it's good sometimes. But do you have any ideas as to why this is happening? I think meta's been getting a lot of hate justifiably because they've made a few different blunders. We mentioned on a previous episode that Zuck has gone on an aqua acquisition spree and barely any of them have paid off. And his most recent, most successful one is now getting unwound and blocked by China. So it's not a good streak for him in general. But I think the

market reaction is specifically because advertising in itself is more of a variable business versus something as hard to find as compute. You have recurring customers with compute, with AWS, with Google Cloud. You have contracts that you sign, long-term contracts with these enterprises and you know they're going to keep coming back. With advertising, it's kind of like, do you think, Google is going to create a better AI search optimization system next quarter, maybe. And I think the metaphor is just oversold at this point. I think people are looking at it as a social media company versus an advertising business. And if I were a meta investor right now, I'd probably be adding onto my position.

Yeah, it may be helpful for people to reconsider these companies, because a lot of the times, the way that they interface with these companies is through the consumer layer. It's like on Amazon, you're interfacing with their Amazon mobile app. On meta, you're interfacing with Instagram or Facebook. but the reality is that there's huge businesses built on the backs of these that is far larger than the actual core business that you're interfacing with on a daily basis, which is interesting. Now, Meadow was perhaps the biggest loser today just in terms of price performance. The second biggest loser was Microsoft.

Now, Microsoft should surprise no one. They're having a little bit of a difficult time when it comes to kind of being at the forefront of AI, but not in terms of their cloud or their infrastructure. Because like everyone else, they have more demand than they know what to do with. they have incredibly high margins, and they're doing really well, pretty much across the board. They generated great numbers. One of the things that I think surprised investors, perhaps the reason why it's down say is because this CAPEX number, it grew to $190 billion for the calendar year of 26, and that includes 25% boost from previous estimates because of this higher component pricing.

Now, for reference, for framing, that's more than the GDP of Greece. It's more in the annual cap-ex of Boeing, Lockheed Martin, and General Motors combined, basically the entire defense industry. So it's a tremendous amount of spending on a business that I think a lot of people are a little confused on. They had some issues recently with the Microsoft and opening I restructuring the partnership agreement that happened earlier this week. It sucks down a little bit, but it seems overall, like as long as Microsoft is leaning into their cloud infrastructure, they're going to be okay. Yeah, I think I relatively agree. Okay, so the reason why I say relatively, they're cloud business, Azure is a beast. It's up 40%. They're printing money. They're just not printing as much

money versus their competitors, which happened to release their same earnings report within the same 90-minute window. So AWS blew them out the water, and they are like the biggest car business. So if you're a smaller car business such as Microsoft Azure, you kind of want to be putting up bigger percentage increases. This wasn't the case. Google Cloud also crushed them as well. So comparatively, if you were looking at all three businesses, even though, Microsoft's statistical increase is better, you're probably going to want to buy Amazon and Google over this, but that might also be an opportunity for you.

The other reason, and I want to put my bare hat on for a second with Microsoft, Josh, what's the name of the AI leader at Microsoft? Copilot? No, I mean the person that leads the AI team. Oh, I have no idea. Exactly. That's my point. Not a clue.

So the leader of AI at Microsoft, I believe his name is, Mustafa Salaman just has been sitting at the helm and not really delivering on the products that you would expect and like Anthropic Open AI and even Google have been absolutely running circles around them, right? So remember, Microsoft co-pilot and the Open AI partnership were the earliest partnerships that you could make.

Microsoft, best investor in the room here, put in $13 billion into Open AI and it is currently worth $132 billion. $27% state in Open AI. And even with that, what is it, a year and a half lead across any of the competitors, they still couldn't figure out how to make co-pilot a really good AI assistant. And the adoption rate for co-pilot across all their enterprise customers, which, by the way, they already have in a contract.

People just didn't want to use it. They requested Claude. They requested chat GPT directly. And you know what Microsoft has done in response? What Mustafa Salomon has done in response? He's created Microsoft. features that are backed by Claude and ChatGPT. And now their partnership is broken. Microsoft still has control over the IP, but what happens when that IP runs out in, what is it, three, four years

time? What are they going to do then? They won't be a forefront. I think people are projecting that bear case in the future and they're deciding just to go with a safer option, which is Google or Amazon. That seems right. Yeah, Microsoft, they've really just kind of missed when it comes to adopting AI. There's another earnings report that I'm very excited about that. By the time you're listening to this episode is out, which is Apple's. I think that's going to be interesting to see how they're starting to navigate this new paradigm shift where they're slowly moving into AI. I know WWDC is coming in June. Hopefully we get Apple intelligence. I mean, of all these companies, they're serving the infrastructure, but in terms of where that infrastructure is actually served, that's all on Apple devices. It's

on all the MacBooks. It's on all the iPhones. So that's going to be really interesting to see. I see a hyperscaler cloud growth chart here. Do you want to talk about this for a second? No, I just wanted to like, as we round up this segment, because I know there's like a lot of numbers here, there's a lot of percentages that we're peddling here. This chart kind of lays it out pretty clearly what the thesis is and what has been proven in these earnings reports. AI-related profit is up massively. It largely comes out of infrastructure, margin, growth, and that specifically has come out of cloud businesses specifically. And who owns the biggest cloud businesses? It's Google. It's Amazon. And even though it's the littlest brother, it's also

Microsoft Azure. And this chart, I expect to see go up and to the right over the next couple of years at the very least. And as long as there's free cash flow coming into these businesses, it's fine that they spend it. That money is going to find a home elsewhere, and that's going to improve the whole market. So I think as we wrap this up, you look at the PE ratio is the price to earnings ratio. This is basically how expensive a company is. So price to earnings ratio, money comes in, a stock gets value. This is the multiple at which the premium of how expensive it is. So meta is trading 16 times earnings. That means issuing 16 of 16 times the forward's projected earnings of the company. On a relative basis, that's pretty low. We have a comp for the dot com

the dot com bubble peak. The PE ratio has got fairly high. Microsoft's that 73 times earnings relative to 25 where we are now. Cisco was over 200 times earnings and Yahoo is over 800 times earnings. Now there are some early signs of this, SpaceX being one of them when I think about its PE ratio. I don't know what it is. It's not public. yet, that's going to probably be close to the Microsoft Cisco level. But in terms of these core businesses that are running, they're not priced very highly at all. This isn't bubble territory. They're not going into debt. They are not spending what they can't afford. They are just taking the revenue that they're earning from their customers and they're deploying it to grow their

businesses even larger. So if you're asking, is there going to be a problem? Are there shortcomings in the market coming soon? The answer is, I mean, Maybe not. Things can break. This is subject to change. I know there's a lot of underlying things that may change the narrative here, but as of now, things seem pretty solid. And these earnings reports look awesome. Yeah. My answer to the bubble question is we're not in a bubble right now. As soon as these companies start taking a lot more debt and these P.E ratios skyrocket, then we are. Listen, it might happen suddenly once SpaceX IPOs, once OpenAIPOs and capital gets sucked out of these companies, it remains to be seen. But for now, theoretically, we are not

that. Well, there's one thing. One last thing is like, so the thing that I'm looking at is so long as that incremental dollars spent on compute is still valuable to the customer, then everything works. It's like as long as the end user is finding value from these AI tools and services, and they're willing to pay for that, everything continues to work. It's once that train stops, once Claude starts to become dumb, once we run out of use cases for it, that's when we run into problems. And we haven't got there yet. So I'm feeling optimistic. It's really funny because you mentioned Claude, you mentioned chat, GBT. Those are. are the two products.

Like, if these two companies implode overnight, we are screwed. But the good news is not just regular consumers like you and I, but very high-powered enterprises that generate billions and billions of dollars every single year, find deep value in these products, and they're embedding it in every aspect of their business. So for now, directionally,

we're heading in what I would say is a positive direction for all of this. I think that brings us to the end of the earnings. We have more to cover. Let's switch context a bit. So, Josh, have you heard of something called Project Deal? No, never heard of it. Fill me in. Okay. So our girl, Olivia Moore, does a very good job of summarizing an experiment that the Anthropic team did for one week. They said, we created a classified marketplace for our employees in San Francisco. It's kind of like a Craigslist, but with a twist, all the deals that were conducted by AI models were acting on behalf of employees. So they spun up a bunch of AI agents and you as a human couldn't list

the item that you wanted to sell. You couldn't write the description. You couldn't pitch it. And you as a purchaser couldn't say, oh, I don't like the look of that. It had to be completely run autonomously by Claude AI agents. So they ran this experiment for a week and did the human exchange by the end of it. And it ended up generating $4,000 worth of purchases and sales across the entire platform. And the customer feedback from the employees themselves were, wow, I actually ended up with items that I really wanted. If you were a purchaser, you gave your agent a wish list. And if you were a seller, you just listed an inventory of things that you had in your apartment and the swap was done after a week. I thought this was a pretty cool experiment for autonomously run businesses or

e-commerce specifically that could happen in the future. That's very cool. And also, here's an answer to why the income will not stop anytime soon. Because think about how many tokens this requires in terms of agentic capabilities. It's like, we started with the chatbots. You talk. It spit something out. That was like, okay, but it didn't use that much tokens. Then we had chain of thought where it actually generated tokens in order to create a better answer. And now we have agents that leveraged not only chain of thought, but now are going off and performing all these skill trees, burning tons of tokens. As this type of workflow becomes more powerful and delivers more value, the amount of tokens that required to generate go vertical. And I think that's what we're seeing right now is not only is the demand for AI going up.

The demand for tokens is going astronomical with it because of these workflows like this one, where I'm sure. In order to get this done at this quality, you consumed a tremendous amount of tokens. And I think, again, this is probably another bullish use case. Now, we have more news as it relates to Cursor. Curser releases their agent harness to the public. What's going on with Cursor? Yeah, okay. So I need to pause for a second. I need some humble pie because exactly a week and a half ago, I was crapping on Cursor. I basically said that they have no moat. They are, and I quote, an AI wrapper that sits on top of the smarter models like Claude and Chat GPT, And, you know, Anthropic and Open Air

can basically eat their lunch at any given moment. I was wrong. So Cursor this week released a really cool product, which was their AI agent harness. Now, for those of you who don't know what an AI agent harness is, you have the main model, but the model in itself is good,

but if you can mold it in a very specific way around a very particular use case, so that includes a bunch of very heavily detailed prompts, it includes setting up the right developer environment. And it includes plugging the right APIs and orchestrating all these things in a very meticulous way. That's known as the harness.

That has grown to be as valuable as the model itself. In fact, there was an eval board that was put out this week because we had chat chip D5.5 come out and we had Claude Opus 4.7 come out. They tried adding the harness from cursor on top of these models. And guess what?

They performed better than the actual models themselves. So now companies are like, well, I want this harness as well. And in fact, Sam Altman himself said this week that the harness in the model are equivocal. They're the same thing and are as valuable as each other. So Cursor decided to do the insane thing, which is release their harness as an API and charge people to sit on top of that. So in an instant, overnight, they created this moat that sits on top of these models and they can now charge a premium for. It was just a genius move by Cursa.

I'm kind of embarrassed that I didn't see it play out in this way myself. but they basically have a product that they can sell that makes chat GBT and Claude 10x better than it already is today. And that's pretty cool. Well, you know who saw that is Elon and SpaceXXAI, which sounds like one of his children's names. It's a mouthful, but they acquired the option to purchase cursor, what last week, two weeks ago? And imagine if that model wasn't anthropic, it wasn't open AI, but it was a Groch model. And now it can be 10 to 20 times better in a harness run by cursor. I think that's probably what they're guiding towards.

So Cursor seems now to be a very important company, a very important puzzle piece, and hopefully GROC can take advantage of that when they either work together or come to the acquisition later this year. Now, there is more news on the front of AI generated content, this one being 11 labs. Now, 11 labs for those who don't know, is the kind of frontier leading edge of audio AI. So that's things like AI voices, voice cloning, and just recently music production. So they have a music platform website in which you can actually sign your voice up to be used for music generation and creation. And they announced not only that they created this new platform, but also that they've paid out $11 million to creators through the 11-Leves voice library. So the question I asked myself, and I asked EJS before we recorded is, why have we not signed up our voices for this creator program? Because my God, $11 million is a lot of money.

It seems like a lot of the pushback around music generation, when it comes to AI in particular, is that it's not real and it takes away from real artists. And I think the reality is that artists who publish on Spotify make virtually no money. You get almost $0,000, even if you get a billion plays. Whereas 11 labs, they're paying out $11 million to a very small set of creators because they can actually earn and participate in the upside of this AI generated music. So in addition to the music being incredibly well produced in some cases, it also creates this really interesting incentive economy around submitting your human vocals to be used for this

music production. We have some samples of music. Maybe we should take a sec. Yeah, give it a listen. Gone, Josh. Pick a, pick a track. Let's see. Chrome candy sounds like fun. Okay, here we go. Okay, I do immediately regret that decision. Okay, so out of all of that, I think we picked the most AI generated one. I do immediately regret that decision. I promise you some of these are actually pretty good. The thumbnail was a bit of a giveaway there. But yeah, to your point, there's a few reasons why I like this product. Number one, the biggest pushback from artists specifically against AI,

and rightly so is this is killing our business. We aren't able to make money. You're stealing our IP and just generating profits for free. This platform not only recognizes your IP, but feeds profits and money back to you. Now, typically in the music industry, it's very complicated. You need to have your IP owner, record label, all this kind of stuff, and it gets very contractory, boring, and tedious.

This is one singular platform that is traceable across the technology itself that can license your voice for content that you haven't even created. It could be someone else, you and I, Josh, that come up with an idea to leverage someone's voice and create a bang, a tune that goes on to makes millions of dollars, and you see money come through that, and it's in a transparent way through this platform. So I think although this is a small infrastructure and platform right now, this could grow to be something massive. Imagine if Spotify used this infrastructure, you know, you'd end up with a bunch more happy artists that end up making more money on average. Yeah, it's pretty cool.

I like 11 Labs. I think they have an incredible platform. I like this new take on music generation. I hope it changes the sentiment around AI-produced music, how it can be an actual tool that empowers creator more than it destroys them. We also have more news as it relates to Claude releases. Now, the month of Claude is kind of over. They slowed down on their daily releases, but they did release something just yesterday that we found interesting, which is a new series of connectors, things like Blender, things like Adobe Creative Cloud, Ableton, which is a music production software, Canva, sketchup, a lot of new applications. One of the most noteworthy ones was Blender, which for those who aren't familiar, Blender is a 3D modeling and rendering software that actually takes really complicated files for rendering things like if you ever wanted to produce a product, it creates the 3D renderings of all.

all the intricacies of something like a keyboard. All the keys, all the switches, the housing for it, and now it can work directly with Claude. I was really interested. I wanted to check out the Adobe Creative Cloud because I'm a huge fan of Creative Cloud and do a lot of work in that world. Unfortunately, it's only just Adobe for creativity.

It's a pretty lame app. And then I started playing around with this more. And the reality is that it's very slow when it does all of these things. It takes a long time to think and work. And while it is powerful, it's just painstakingly slow to actually get outputs and engage with these tools right now. But it seems pretty interesting.

I would hate for it to make a mistake on like a 10,000 part rendering of some complicated object and you don't know what it is. So it's high risk. Use at your own discretion. But I think it is an outworthy news for this week at least.

Okay. So for the final topic, Josh, I have a nomination for the worst AI product ever created. And I'm sad to say that it was created by one of the biggest companies that we just spoke about earnings reports. So thank God this isn't the moneymaker. So this is a new feature from Amazon where when you go on their e-commerce business, the Amazon platform where you order toilet paper,

like I mentioned earlier on, you now have a new AI feature which generates a podcast about the products where two AI hosts discuss the product that you're viewing and takes your questions as if it's in a call-in show. So just to be clear, you press the button, right? This is like a demo here. I'm not going to play it. But, like, you know, you're looking at, what is this? HomeCare Pro Rapid Relief Rash Cream. And you can generate two AIs that, like, talk about the product. And if you have a question, you can type in the question and they answer it live. So as you're scrolling through the page, I don't know why you would want to use this. Maybe I'm missing something. We are in the podcast game and even I hate this product with a visceral amount of myself. Josh, can you help me unpack this?

Like, why would I want to use this? Is this even real? This seems like they're trolling. Like, is this even a real feature? I'm like very skeptical. But it's funny. I mean, previously, NoPokeLM, which is Google's program, they did this. They allowed you to kind of insert any sort of research or idea into their program.

And then it would generate you a podcast. This seems to be doing a live call in Q&A based on people's reviews and tough questions. It's, um, I guess if anything, it's a testament to the cost of content coming down to a level that is so marginal, Amazon is looking to implement it basically for free. Is this necessary? Probably not. Are people going to use this? Probably not.

Is this real? I'm not even sure. If anything, it's a really good feeling demo. But yeah, I found it really funny that, I mean, it's just a testament,

again, to the cost of everything going down, the accessibility of AI, and the AIification of everything. Not everything needs to have a podcast. Not everything needs AI. I don't know. Funny sort of end on. But I think that is everything.

This was kind of long. If you made it at the end, you've really, you've made it. You're up to date. You know all of the, things that matter this week. We had a few other episodes this week that were pretty interesting, if you missed. One is on that China deal with meta. One thing that wasn't covered in the earnings report

is this meta and Manus acquisition issue where China actually blocked it. Meanwhile, the CEO of Manus is a vice president of meta and working at the offices currently. So it's a huge disaster. We had an interview earlier this week where we spoke with the guy who is responsible for USVC, an easy way that you can invest in Anthropics, SpaceX, and Open AI before it goes public. That was pretty interesting. And we also covered Sam Albin and Elon Musk's trial. Now, we don't have any updates on that. Today, we're going to be covering those in the coming weeks. But yeah, thank you guys so much for watching. Any final thoughts before we leave them for the weekend? No, I think we've kept them here for a while, but you are officially caught up to date with

everything and anything going on with AI, everything from product releases to earnings reports to drama between Elon Musk and Sam, we do grudges. We do everything on this show. So the point is, we have extensive coverage. And if you aren't a subscriber, if you haven't rated this show, even if you hate us, leave us a comment, give us a rating.

We would love to hear from you. It helps us out massively. And yeah, we will see you next week on the next one. Thanks. It's a great weekend. See you guys. See you.