Loading summary
Transcript49 lines
- [00:00]
A
Welcome back to the Rundown interview edition. Today we are talking to Alex Heath, author of the Sources newsletter and one of the most plugged in reporters in Silicon Valley. This is Alex's second time on the show and he is joining at a wild time for AI. We covered a lot in this conversation, including why Wall street is confused about Meta's AI plans, the approach that Apple is taking to AI, the impact that Apple's lawsuit could have on OpenAI's hardware ambitions, and whether OpenAI Anthropic will IPO. This was a really great conversation. I think you guys are going to really enjoy it. So let's get into it. All right, guys. Today we are joined by Alex Heath. He is the author of the Sources newsletter. He's one of the most plugged in people in Silicon Valley. Alex, welcome back to the Rundown.
- [00:49]
B
Good to be back.
- [00:50]
A
Yeah. Thank you so much for coming back. The last time you were on was almost a year ago. It was October of 2025. And I gotta say, a lot has changed in the world of AI and also with you. You know, you just launched Sources back then and now you're almost a year into this. How is it like, you know, going independent? What was, what's been the biggest surprise to you?
- [01:09]
B
It's been great. It's wildly exceeded my expectations. Yeah, it's been really fun. I've got Sources been doing it for about a year, like you said, having a lot of fun covering the race.
- [01:20]
A
I want to start today's conversation talking about Meta first, because they reported earnings last night. Their stock is getting crushed right now. And I think a lot of it has to do with, you know, Meta just continue to spend a lot of money on capex or free cash flow has pretty much disappeared. And, you know, I listened to the earnings call last night and I feel like Zuck is just not doing a great job explaining what their AI strategy is. What is Zuck's master plan? How are they planning to make up for all this money they're spending on AI? You know, you recently talked to Alexander Wong, who is their head of AI. What is the vibe inside Meta right now?
- [01:58]
B
I think they obviously want the street to understand the strategy. I think the challenge for everyone trying to understand the strategy is, you know, they have a lot of different products and there's a narrative to be told about how they're using AI to do things like replace how the algorithms work in Facebook and Instagram or how ads are targeted across Meta's properties. Right. That's one application of AI. It's less sexy. And I'm not sure investors look at that and think it's worth all the Capex spend. And then you've got the things like the glasses which have AI components but really I would say are behind in terms of, I think where Meta wants it to be on the AI stuff. And then the Meta AI assistant still has not, you know, gained meaningful consumer market share relative to, you know, Gemini chatgpt, Claude. So they have a challenging story, I would say on like the consumer assistant agent front. I noticed in the call Zuckerberg was talking about, he has this, you know, vision for everyone having their own kind of personal assistants and agents that do things for them for days at a time autonomously. I think we're pretty long way off from that. Working at scale is my read of where the technology is and talking to the people who build it. But Meta's made some interesting acquisitions. Aqua hires, they certainly have talent in house and they do have compute, as was also a theme on the earnings call, maybe too much compute. So I think they have to build a product that people love that isn't just something like Meta AI that is being put in places across their apps and that juices usage because you see it and you run into it by accident and Instagram or WhatsApp or whatever. So I think Meta AI, it's still kind of unclear its value, its long term value as an AI product relative to all the other things and they have to figure that out.
- [03:54]
A
Yeah, I think obviously like AI is probably helping Meta improve their, their ad platform. Their revenue growth is pretty phenomenal. But yeah, it's just not enough. If you're dropping 140, $150 billion on CapEx. I think the street wants to see something else that's going to pay off for that investment. Now it could be this cloud initiative that they've talked about that has been reported on and Zuck kind of hinted at, but there's no details on it, there's no customers, there's no timeline to sell all this excess compute. Is there any traction there right now? We didn't much in the earnings call. What are you hearing on the inside?
- [04:30]
B
The COMPUTE strategy is still being figured out. You know, I think when you buy this much compute, ideally you're using it for your own demand, your own inference, your own training. Meta doesn't have the inference demand. I think that it may be hoped for its AI products. They have, you know, many of the largest products in the world with their existing products. So I think they'll use that compute, no matter what, if they needed to. But you can get a significant premium as Zuckerberg was talking about and Elon Musk has showed with Colossus. You can get a significant premium right now if you have available, wired up ready to go compute in the market. And maybe that premium is a higher ROI than, you know, a free meta AI product that actually costs money on inference to serve that you're not monetizing. Maybe it's a better business decision, especially when you spent so much money on your build out to rent some capacity in the short term until you're, you know, you have champagne problems and you have such a great monetizable AI product that you need to buy more compute for it for yourself. Which Meta, I don't think is in that position yet. I do think they'll get there like I do think they will be a hyperscaler at some point. I just don't. I think it takes time. I mean, you look at Google Cloud, it took Google, you know, over 10, you know, years really to build that business. The reliability, the uptime that people, you know, expect for a hyperscaler. I've seen people talking about, you know, how it's easy to get to a decent amount of uptime, but, and reliability. But to get to that last, you know, 99.999% reliability as a hyperscaler takes billions and billions of dollars and years of experience and Meta's just getting started. But you know, I, I do think they, they have compute, which is, you know, it's a good strategic advantage right now and everyone's looking for it. And I think Elon's shown that they just have to figure out what to do with it now.
- [06:25]
A
Yeah, I mean, I think the Google example is great. It took Google Cloud so long to build up just like the manpower, right? You have to build your sales organization. You got to build all these different parts of the business in order to be a true hyperscaler. And they finally started doing that and they're seeing the payoff now. But it might take Meta, like you said, five, six, seven years, if that, to really get there. And I think there's like this element that, well, is, is Zuck really serious about this? Is this another metaphor metaverse situation 2.0 where they're just throwing stuff at the wall or hoping that they kind of land on something, but there's no like strategy moving forward. And I think that's what is causing the uncertainty right now with the stock price.
- [07:04]
B
I think you need to also be at the frontier on models for the amount of money that matters, spending on AI and AI research. So I think the fact that he had, you know, maybe the most well publicized, you know, AI lab build out of all time with all the hiring he was doing and you mentioned Alex Wang, you know, folks like that, I think that put a lot of pressure on the strategy and you know, Muse Spark, their model is, is good. You know, it's not pushing the state of the art, but it's right up there. And I think they're catching up rather quickly. I mean, I think if they come out with a model in the fall that quite, you know, measurably pushes the frontier forward on key benchmarks, like maybe the stock gets rerated. I think, you know, there's a sense of like he who, he who has the frontier models, you know, controls his destiny in this market and has the mandate of heaven. And you see that with, you know, anthropic and open AI and, and Google DeepMind up until recently. Right now the question is, can Google DeepMind get back to the frontier? And you see that kind of weighing on the stock. So a lot of pressure on Wang and his group to ship as well.
- [08:08]
A
Well, a company that's not doing anything when it comes to frontier AI is Apple. So I want to talk about them. They're actually reporting earnings this afternoon after the bell. So by the time people listen to this interview, it'll have already come out. I mean you can kind of see how the vibe has shifted when it comes to like how the market is treating AI. A year ago they were kind of the laughing stock as they didn't really have an AI strategy. A lot of internal turmoil as well. And now the stock is hitting all time highs, you know, across $5 trillion in market cap for the first time ever. Do you think that Apple is like a genius or just lucky? Because they're just pretty much, they're pretty much counting on Google now to kind of supply all the AI tech for this upcoming series that they have.
- [08:47]
B
The new series, really good. I have it on my phone. I do think that's a winning product for them and I think they are feeling very good about that in the Google partnership. You know, people maybe forget they've been deep partners with Google on search for, you know, going on over 15 years at this point. These are two companies that know how to do these kind of partnerships together. They trust each other. They've been through fire together with the antitrust lawsuits on search, for example. I think Apple's very comfortable being close to Google on AI for the foreseeable future. I think those are two companies that have their own swim lanes and have almost like a pact to not, you know, directly go into each other's territory. At least so far. I think people see Apple's, you know, devices and their, their expert edge on silicon for their devices as, you know, maybe with the rise of open AI and open weight models and people running things locally and wanting to control their models, Apple devices are the best for that. We saw that with that open claw early hype where Mac Minis were getting sold out everywhere. They're a beneficiary of open models. And every time I get asked about Apple I just say, well, is there anything happening in AI that's going to make you not buy an iPhone in the fall? So far that continues to be true, that nothing happening would keep someone from buying an iPhone. Right. There's no magical AI device that someone's made that will replace the iPhone. I don't see that happening anytime soon. And as long as that remains the case and now they've got the new Siri actually working, I think Apple's going to be just fine. I think if anything they have a better profile because they haven't spent tens of billions of dollars on AI without anything material to show for it like some of the other players we've been talking about.
- [10:44]
A
Yeah, the only impact that Apple's having when it comes to AI is like the rise of memory prices which is forcing them to raise the prices on their device. Now we'll have to see if that's going to have a meaningful impact on their sales upcoming because if the new iPhone's a couple hundred dollars more than the previous one, people might choose to hang on to their old phone instead of upgrade.
- [11:03]
B
Yeah, I think we'll see. I mean they're going to have the foldable based on all the reporting I've seen the foldable iPhone, which will be the most expensive yet. I think early adopters, folks like us will still get that. Maybe there will be some pressure on them from pricing, but so far it doesn't seem like they've shown it.
- [11:20]
A
Well, I want to talk about their Apple's previous partner, OpenAI. I mean that is crazy how fast that relationship turned. They're like going from partners at the WWDC in 2024 to now. There's lawsuits, there's like, you know, allegations of trade secret theft when it comes to hardware. Do you think this is actually going to slow down OpenAI's hardware ambitions? Because everything that I've seen they seem to be very confident that, like, what they have, what they're working on right now is, is going to be pretty special. So what have you heard from. From the OpenAI camp?
- [11:52]
B
I mean, their statement is, you know, that they don't need Apple secrets. I think the lawsuit's pretty bad. I think it's pretty bad. I think, yeah, it's. There's a lot of very specific things in that lawsuit that Apple is alleging that they've done, including their leadership on the hardware team. And will Apple win? I don't know. I'm not a legal expert. Do I think this case continues and gets maybe messier? Probably. Does it delay OpenAI's planned kind of fall launch of their first device, which everyone who hears about these things believes to be kind of an AI speaker puck like device? I don't know. I think it probably still gets announced unless Apple wins some kind of injunction. But Apple's going for the jugular here. I mean, I think they feel, they feel very personal about this and it's certainly not a good look when it's, you know, before you've even shipped your first device, you're getting sued for trade theft from what was one of your largest partners. I mean, ChatGPT is still integrated into iOS. It's severely downgraded in terms of its prevalence and ease of use in iOS with the new version that I have on my phone and the new Siri. So I think that partnership will go away. And, you know, OpenAI at one point thought Apple was going to be kind of how Google and Apple are now. You know, it's going to be like a huge partner and distribution source for them. And that turned out obviously to not be the case. So I think it's messy, man. I don't think it's. I don't think it's clear what's going to happen, though.
- [13:25]
A
You make a good point about the lawsuit. I wonder if it's just, does Apple really feel that threatened or is it more just like a personal thing where Apple is like, has never really faced any real competition when it comes to hardware. And now you have John Ternus who's taken over as CEO. He's a hardware guy, so a lot of the people that OpenAI poached were his people. So I wonder if there's just some bad blood right now and John Turner might be leading the train here. It's more just. It's more like drama than anything. And I do wonder if it's going to ultimately not be that Big of a deal when it comes to OpenAI's hardware ambitions.
- [13:57]
B
It may not derail the first device. Does it derail their hiring and momentum and internal momentum? Probably. But yeah, I mean I don't to your point about Turnus, like I doubt he was personally driving this lawsuit. It probably was a board level thing for Apple. Apple's very litigious about this stuff. They sue ex employees all the time. There are the famous like Steve Jobs emails about poaching back in the day Adobe. This is not like out of, out of character for them. So I wouldn't say it is personal. It's obviously personal. These are a lot of people that OpenAI hired, you know, used to work at Apple and worked in senior positions, including Johnny I've. Right. Who's kind of running that project with Sam Altman at OpenAI. But yeah, I don't think, I don't think Apple is doing this just to kind of play around. I think they're serious about it.
- [14:52]
A
Yeah, yeah. I mean I'm really looking forward to seeing what this new device is. I think like the headline wasn't super sexy. It's like it's a speaker puck AI thing and it's like, well, are they just making like a, an Amazon Echo kind of deal? But the more I read into it, the more online, especially their new voice models that they just released, I mean there's some interesting tech happening here. So I'm really looking forward to seeing what open AI has because, you know, I feel like this summer has been big for OpenAI. They kind of lost a lot of their juice earlier in the year to Anthropic. Anthropic just had all the hype and now OpenAI is starting to make some moves again with their super app that they launched a couple weeks ago and then obviously 5.6 soul. So this could be what gets open AI kind of back on top when it comes to like just all the hype in the, in the, in the AI space.
- [15:37]
B
Yeah, it could be. I think the main thing for them is continuing to have the best models and, and ChatGPT continuing to grow, which its growth has slowed a lot, which is an issue. But I think they feel really good about Codex in the API business. I saw some leak yesterday that their CFO said I think in July they added like more net new ARR than all of Q2. So yeah, they're kind of, they're hockey sticking again and maybe OpenAI is back. You know, these, these AI race, these Frontier AI lab, you know, narratives and vibes tend to be on these kind of six to eight month cycles where someone's, you know, it's, it's like the so over so back meme, you know, that's like, that's like covering the AI race every day.
- [16:18]
A
Do you think that, how real do you think is this threat from, from China right now? You know, a couple weeks ago we had Kimmy K3 open weight models that are, you know, frontier level when it comes to performance at half the cost of what Anthropic and OpenAI charge for their models. And now we have Jensen come out last week endorsing open weight models because obviously that's good for his business because he wants as many open weight models there are. He doesn't want just the AI industry to be dominated by OpenAI and Anthropic. Do you see this as a real threat? Like are people taking these models more seriously? And inside, inside Silicon Valley?
- [16:54]
B
Oh yeah, for sure. I mean, yeah, these models are good, they're way cheaper. So it's going to put pricing pressure on the closed labs. It already is. I think the cat's out of the bag here. I think China is catching up. I'm not a politics expert. Seems complicated for, you know, geopolitical situation that we have with them, but certainly from a tech perspective the models are very good. Moonshot's good, you know, with Kimmy. And I think everyone expects that to continue barring some kind of intervention from the government, which some players would probably like to see happen selfishly, but understand that it's bad for macro business and the markets to start intervening like things like that. So I don't think anyone leading these labs like privately feels like that's a good idea even though they kind of want it to happen for their business. So yeah, I don't know. I think it's, I mean my, my view of it is it's probably a good thing, right, that there's competition from other countries and we don't feel like we're just totally crushing and there's no one close to us. Competition tends to be good. Now if you believe recursive self improvement is coming and AGI is coming and you know the world is going to end because of these models, then you're like, no, China, you know, needs to be slowed down. I just haven't seen evidence yet that we're on the precipice of some kind of God like change in, in these models that, you know, totally changes the fabric of society overnight. I just don't think the world works
- [18:32]
A
like that, yeah, same here. I think all the AI God stuff is, is a little bit, a little bit too woo woo for me. But I do like the fact that yeah, we have Chinese models to you know, put pricing pressure on OpenAI anthropic and hopefully, you know, it keeps pushing the prices lower and lower. The market is kind of going through this weird phase right now with AI stocks just having a very tumultuous time. Is the market going to reprice what these frontier labs are worth? I'm seeing more and more commentary about, well, is OpenAI and Anthropic worth a trillion dollars? Are they going to be able to capture all the value in the value chain if we have a world where open source open weight models are dominating And I Wonder if like OpenAI and Anthropic might have missed their IPO window as well. When you were on last time, a year ago, approximately a year ago, I asked you if OpenAI and Anthropic would IPO before 1-1-2028. You said yes. Do you still feel that way? And I mean do you think they'd fetch a trillion dollar valuation?
- [19:32]
B
To be honest, if I knew the answer to that question, I wouldn't be on this podcast. That's true, you don't have to answer
- [19:37]
A
the trillion dollar part. But like do you think the IPO is still gonna happen?
- [19:40]
B
Yeah, for sure. Anthropic will probably go out this fall, probably October ish. If I had to guess. I don't think OpenAI is ready. I think OpenAI may be early next year if they can. Obviously this changes week to week based on the markets, everything, competition, you know, how SpaceX does all these things factor in to these companies decisions to go out. But I do think your point is right, that like maybe the peak hype moment to go out was, was a little while ago. I mean the, the word is like Anthropic will probably exit this year doing 100 billion plus in ARR. I'm not a math guy. I don't know what the multiple to get to a trillion on that is. I think it's like it's not, it's not that crazy of a multiple actually. So I think Anthropic, the more it scales, it's a more attractive looking business than Open AIs because OpenAI, the vast majority of OpenAI's inference, aka ChatGPT is losing money because most ChatGPT users don't pay to subscribe and the ads are not scaled yet. Right. So chatgpt as it scales loses money. Anthropic as it scales, its margins get better until they can fix ads. And the API business for them is apparently hockey sticking now. And Codex is growing, so maybe there is more complexity to OpenAI. They have more levers. And in a secular downturn, would you rather be like an enterprise API business or something with a billion plus consumers that can weather a storm? I don't know. I think Google and Meta have shown us that having a billion plus user platform gives you a lot of longevity in the world. So OpenAI probably has that as well. They have that to their advantage. But in terms of timing, I think it's anthropic first and, and OpenAI maybe early next year if they can. I think they have to get their, their house in order a little bit more and the story needs to be there and you know, they need all the numbers up to the right and they have some numbers up to the right. But again ChatGPT has been a, a sore spot in terms of growth this year for them. So Anthropic is more like everything's up to the right. So that's also I think, motivate. I mean there's also just the rivalry of like who goes first, who gets the retail dollars first, who gets the population. And you know, people definitely saw what Elon did and, and now in the way that SpaceX is traded down, it's like, okay, does that factor into, right, how you price and can they go out at a trillion? These are all like, you know, there's bankers making a lot more than us that are trying to figure this out and they, they don't know, they don't know right now. So like I don't think anyone knows. But this also the midterms, like it's going to be an interesting fall like regardless of what happens. So yeah, I think the jury's out on all of that.
- [22:33]
A
Yeah, I think it's great for ours, our business. Right. We have a lot to talk about.
- [22:36]
B
Yeah, exactly.
- [22:36]
A
About.
- [22:37]
B
Yeah.
- [22:37]
A
But I think the SpaceX is performance and how fast, like I mean it's, it's not doing great right now and like it's, people are asking questions like if it was SpaceX potentially the top. And then now with SK Hynix as well. They did their US listings earlier in July and now their stock's not doing so great and did OpenAI and anthropic. Mr. Window, I can't wait to talk about it for the rest of the year. I'm gonna, I'm gonna ask you one more question before we Hop off. It's a hot take of a question. What do you think about the chatter that both Sam Altman and Dario at Anthropic. So Sam Altman at OpenAI and Dario at Anthropic are going to be replaced before these companies ipl. I've been hearing some chatter. Maybe it's just my group chats, but like there's hot takes being thrown around that both of them will be replaced before. These are like grown up publicly traded companies.
- [23:26]
B
I don't think either of those guys are going anywhere.
- [23:28]
A
Okay.
- [23:29]
B
I think I could see Sam getting like kicked up to chairman and them finding a new CEO at some point. I think he would probably like that. And whether that happens before the ipo, I'm not, I don't know. Dario I don't think is going anywhere. And Anthropic is like, is a brain trust of I think around eight co founders who all have equal voting stock and decide on things together. And Dario is like very, very ingrained in that place and the face of the company. So I don't. And Anthropic is doing okay. So I don't think, I don't actually see either of them leaving. But again, maybe, maybe if I were having a hot take it would be like Sam is chairman, not CEO at some point. More to come on that and more to come.
- [24:14]
A
Well, I can't wait to, you know, read your, read your newsletters on this stuff. Alex, thanks again for hopping on and give. Give the people, you know a few places where they can read your stuff.
- [24:21]
B
Just sources, news.
- [24:22]
A
We'll put the link in the description for sources and thanks for hopping on and we'll have to bring you on again in a few months to kind of recap some of this stuff.
- [24:29]
B
Sounds good. Thanks for having me.
- [24:30]
A
Appreciate it Alex. Well, all right guys, hope you enjoyed that conversation with Alex Heath. I always enjoy talking to him and getting his perspective because he's actually talking to the people running these companies. Let me know in the comments of what you guys think. Do you think that Meta will ever figure out their AI plans? And do you think that opening in Anthropic will IPO sometime this year? And do you think that Sam or Dario could be replaced as CEOs? Drop your thoughts on Spotify and YouTube and while you're at it, consider giving us a five star rating or thumbs up as well. All that engagement really does help us out and it helps other people find the show. Thank you guys so much for listening, watching and commenting. Shout out to Mike for all the work behind the scenes and we'll see you guys back here on Monday.
- [25:17]
B
I see you.
- [25:18]
A
Avatar Fire and Ash is now streaming on Disney plus. It's the film critics are calling the best avatar yet. A true epic and completely jaw dropping. This is the only pure thing in this world. Return to Pandora on Disney plus. It will be an adventure for the whole fam and watch the Oscar winning phenomenon at home.
- [25:38]
B
This is sick.
- [25:42]
A
Avatar Fire and Ash now streaming on Disney. Rated PG 13. Close your eyes, exhale. Feel your body relax and let go of whatever you're carrying today.
- [25:54]
B
Well, I'm letting go of the worry
- [25:56]
A
that I wouldn't get my new contacts
- [25:57]
B
in time for this class. I got them delivered free from 1-800-contacts. Oh my gosh.
- [26:01]
A
They're so fast. And breath.
- [26:03]
B
Oh, sorry. I almost couldn't breathe when I saw the discount they gave me on my first order. Oh, sorry.
- [26:10]
A
Namaste. Visit 1-800contacts.com today to save on your first order. 1-800-contacts.