In short
U.S. government unblocks Anthropic’s Claude “Mythos” for 100+ institutions; OpenAI previews GPT-5.6 (Terra, Luna) with limited access; debate over government “winners/losers” in frontier AI; customers cut AI costs via cheaper model routing; billing overcharges; Apple raises Mac/iPad prices 15–30% due to memory/storage costs.
Guests
Ranjan Roy of Margins (podcast co-host). No other guest is interviewed in this transcript.
Key claims
- Mythos access expands after a U.S. de-escalation; “Fable” (consumer-facing weaker model) may also be released later.
- OpenAI’s GPT-5.6 rollout is restricted to “trusted partners” pre-approved by government.
- Regulation may push firms toward open source and reduce frontier-model ROI; could create an “invite-only” upper class.
- Enterprises are actively switching to cheaper OpenAI/Anthropic models to control bills; some report large annual savings.
- AI billing can be opaque; an auditing startup found $1.7M mistaken overcharges in $34M audited.
- Apple’s price hikes are framed as greed despite memory cost increases.
Notable examples
- Ensemble Health Partners: switched to a model priced 1/20 as high; saves nearly $700k/year on 15,000 appeal letters/month.
- Vaudit: audited 60 companies; $1.7M overcharges (Mar–Jun) mostly for Anthropic Cloud Code.
- Apple: MacBook Air +$200 to $1,299; iPad Pro +$200 to $1,199; iPad Air +$150 to $749.
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Chapters
Tap a time to open that second in VOEpisode Introduction and Guest Introduction
0:00 to 0:27
Hosts introduce the show, today's topics, and guest Ranjan Roy.
“In the face of ongoing disruption and opportunity, TMT leaders need to deliver tangible results, not just ideas.”
Episode Introduction and Guest Introduction
0:45 to 1:36
Hosts introduce the show, today's topics, and guest Ranjan Roy.
“Welcome to Big Technology Podcast Friday edition where we break down the news in our traditional, cool-headed, and nuanced format.”
Anthropic's Mythos Release
1:36 to 3:59
Discussion on the release of Anthropic's Mythos and its implications.
“Are you on the Mythos list out of the 100 U.S.”
OpenAI's GPT 5.6 and Its Naming
3:59 to 6:36
Hosts discuss OpenAI's new models and critique their branding choices.
“Well, it does seem basically, you know, when you take it in combination with this week's OpenAI news that the government has installed itself as the approval of frontier models.”
Regulatory Landscape and Market Control
6:36 to 11:22
Exploring the implications of government regulation on AI model access.
“a non-cryptocurrency crowd and really try to get to something a little bit more memorable.”
Open Source vs. Frontier Models
11:22 to 14:00
Discussion on the shift towards open source models in AI development.
“So it's just kind of taking the same approach to Frontier Labs in my mind.”
Open Source Models and Their Implications
14:00 to 16:41
Explore the discussion on the implications of banning open-source models and their cost-effectiveness compared to frontier models.
“And that's why you have people talking this week about banning open source models, which I don't think would be good either.”
Business Models and IPO Implications
16:41 to 18:59
Delve into how emerging business models and IPO expectations affect the AI industry, especially concerning OpenAI's strategies.
“All the ARR growth that was shown to the world from, call it September-ish, October-ish to February, March, was people just cranking on whatever the latest, most expensive frontier model was.”
Distillation and Competitive Strategies
18:59 to 20:55
Analyze the strategies surrounding model distillation and how companies like Anthropic are handling competition and regulatory challenges.
“about as these frontier models actually, you know, like have widespread adoption, that's where the distilling starts to take place.”
GPT 5.6 Efficiency Compared to Mythos
20:55 to 22:30
Learn about the performance and efficiency of GPT 5.6 in comparison to Mythos and its implications for cybersecurity.
“You say you can cure cancer in a few years.”
Show all 21 chapters
IPO Speculations and Market Trends
22:30 to 24:38
Discuss the speculations surrounding OpenAI's IPO timeline and the shifting market dynamics affecting AI valuations.
“Yeah, I mean, you see price constantly being now a highlight rather than kind of some afterthought.”
Shifts in AI Market Dynamics
24:38 to 28:00
Examine the recent shifts in AI market dynamics, including changing attitudes towards frontier models and operational efficiency.
“it looks like they are going to potentially postpone.”
Economic Headwinds for AI Companies
28:00 to 30:00
The discussion centers on the challenges AI companies face as customers seek cost efficiencies.
“To me, the curve is no longer up and to the right from where it was a few months ago.”
Customer Strategies for Cost Reduction
30:00 to 32:10
Exploration of how companies are adapting by switching to cheaper AI models.
“And we're back here on Big Technology Podcast with Ranjan Roy of Margins.”
The Impact of Pricing on AI Business Narratives
32:10 to 36:40
Discussion on how rising costs and pricing strategies affect AI companies' growth narratives.
“I mean, to me, the most interesting part of what you just said is you called it an economic headwind.”
Billing Inaccuracies in AI Services
36:40 to 40:00
An examination of billing errors reported by customers and their implications for AI companies.
“And they should be trying to lead that and they're not because it's not good for them.”
Apple's Price Increases and Market Dynamics
40:00 to 42:08
Discussion on Apple's recent price hikes on Macs and iPads due to rising component costs.
“I mean, it shed about, I think it was down 15 % or so on the week.”
Apple's Price Increases Explained
42:08 to 45:55
Discussion on Apple's recent price hikes for Macs and iPads and the underlying factors.
“And then it's like surpassing Amazon, surpassing Meta, closing in on Microsoft.”
The Implications of Apple's Pricing Strategy
45:55 to 51:44
The hosts analyze the potential impact of high prices on Apple's business and customer relations.
“Yeah, but we're all locked in and can't escape the Apple ecosystem.”
In Memory of Om: A Reflection
56:00 to 56:45
The hosts honor the legacy of Om and his impact on the tech industry.
“lessons a guiding light for the future with much love and prayers for your family.”
In Memory of Om: A Reflection
57:28 to 57:39
The hosts honor the legacy of Om and his impact on the tech industry.
“Requires Google account, Google Health app, internet, and Google Health premium subscription.”
Transcript
Automatic transcript. May contain errors.0:00Big Technology Podcast Host:In the face of ongoing disruption and opportunity, TMT leaders need to deliver tangible results, not just ideas. When pace and performance matter most, PwC combines market insights and deep sector experience with AI, cloud, and emerging tech to accelerate your transformation and drive measurable ROI from strategy to execution. PwC can help you anticipate what's next, outpace disruption, and compete. For more information, visit pwc.com. Mythos is back and OpenAI's most powerful model, GPT 5.6, is out too. But will the public ever get to see either of them? And Apple hikes prices massively to keep pace with memory prices.
0:43Big Technology Podcast Host:Is it greed? That's coming up on a Big Technology Podcast Friday edition right after this. Welcome to Big Technology Podcast Friday edition where we break down the news in our traditional, cool-headed, and nuanced format. Matt, we have a great show for you today, coming a bit later than usual, but we are here on the feed. We're going to talk about the return of Anthropics Mythos, the debut of OpenAI's GPT 5.6. OpenAI, by the way, may be delaying its IPO to 2027, or maybe it won't. We'll discuss that. And Apple hikes the prices up on MacBooks and other devices by anywhere between the teens and 30%.
1:20Big Technology Podcast Host:So we'll talk about what's behind that, the memory crunch that's underlying a lot of the price increases you might see in consumer electronics, and whether it is unavoidable or, as some might say, greed. Joining us as always to do it is Ranjan Roy of Margins. Ranjan, great to see you.
1:37Ranjan Roy:Great to see you, Alex. Are you on the Mythos list out of the 100 U.S. institutions, major companies, and government agencies that now have access to Mythos? I am not.
1:48Big Technology Podcast Host:If the government is picking winners and losers, then I am without a doubt a loser. So let's go through the headline here. It's something that we have breaking as we speak. The U.S. releases powerful anthropic model Mythos to some U.S. companies from Semaphore. The U.S. government Friday lifted its block on Anthropics' powerful Claude Mythos 5 AI model, allowing the company to release it to more than 100 U.S. institutions, including major companies and government agencies. The decision in a Friday letter sent to Anthropic is a major de-escalation in the confrontation between the Trump administration and one of the world's most valuable private company.
2:26Big Technology Podcast Host:The letter is silent on Fable 5, a weaker version of Mythos that was briefly the most powerful AI model available to consumers. People close to the talks said they are moving toward releasing Fable as well, though the timeline is unclear. Howard Lutnick, the Commerce Secretary, has given his stamp of approval. Well, Ranjan, it looks like this long outage of Fable is about to end and so ends the saga. I don't know if we have any details about whether there have been actual structural fixes or whether the guardrails or the anti-jailbreaking guardrails that the U.S. government has asked for are there.
3:07Big Technology Podcast Host:What do you make of the news?
3:08Ranjan Roy:it's for me the most difficult part of trying to process all this is i mean more on the overall regulatory landscape of who has control of what and when i think uh i mean we've talked about is mythos more marketing and fear-mongering or is it a true danger to kind of the global structural order. And I don't know, it's like, do you feel that the government is actually has any real plan around how they're dealing with the anthropic now? Like, it looks like this just feels like an extension of what originally Project Glasswing was, whatever that was, 20 to 30 companies, now we're at 100. But I don't really understand where this is going, or what's fundamentally changed.
3:58Right.
3:59Big Technology Podcast Host:Well, it does seem basically, you know, when you take it in combination with this week's OpenAI news that the government has installed itself as the approval of frontier models. And I think we should talk about the implications there. Let's get into the OpenAI news because I feel like it really couples together with the anthropic news. So this is from a blog post from OpenAI previewing GPT 5.6 Sol, a next generation model. Today, we are beginning a limited preview of the GPT 5.6 Series Solar flagship model, Terra, a balanced model for efficient, high-volume work, and Luna, a fast, affordable model for everyday work.
4:34Big Technology Podcast Host:I'm just going to pause here for a moment. They're calling their two of their three models, Terra and Luna. Isn't that the worst branding mistake ever made?
4:44Ranjan Roy:Thank you. Thank you. That was the first thing that just jumped out dramatically to me. For listeners who don't remember, those were the names of cryptocurrencies that crashed dramatically. One by 99.999%, the other by Terra by 98%. And Terra was a stable coin as well. these were kind of the poster children for the crypto fallout from a couple years ago so like why i mean i'll take that over spud i guess their code name but how do you think they came up
5:19Big Technology Podcast Host:with those names all of them actually i think they were trying to find some version of like the anthrop anthropics all or like a version of a story like a sonnet and an opus and a fable so they were trying to make their own with these earth and space celestial nicknames like terra luna and soul uh they're obviously like they they have are wide open to the fact that anthropic has beat them on branding but i would just say for goodness sakes don't use terra and luna read the room guys the room my god please and it's not only that they crashed it's that and And correct me if I'm wrong here, it was a scam, right?
5:59Big Technology Podcast Host:It was a scam that basically got everybody to invest in these two thinking that they were stable and people lost their shirts. That's why I would stay away from that naming. Yeah.
6:10Ranjan Roy:Actually, now that you did explain it to me, it didn't jump out at me, the whole kind of like earth, wind, fire, space type stuff. maybe I am feeling it a little bit separating out the cryptocurrency association of actual full-scale fraudulent scams that had a lot of people lose a lot of money. Maybe I am liking it a little. I don't know. Maybe I'm liking a little bit. Maybe we're trying to appeal to a non-cryptocurrency crowd and really try to get to something a little bit more memorable. I don't know. I've just gone to undecided from net negative on these names.
6:51Big Technology Podcast Host:Well, I'm still against, but I suppose it's better than harness. Okay, so let's talk about, let's get back to our main story here, which is that here's what OpenAI says in its blog post that the release of this model is going to be restricted at first. We believe in broad access, but we are starting with a limited preview for a small group of trusted partners whose participation has been shared with the government before releasing more broadly. During this preview, we will continue testing and coordinating closely with the partners as we work toward broader availability. We don't believe this kind of government access process should become the long-term default.
7:27Big Technology Podcast Host:So obviously, they saw what happened with Mythos and Fable. They don't want a repeat of that. So they're kind of letting the government handpick the customers that they're able to work with and those they can't. We're going to get into the model's capabilities because it does seem pretty impressive. but I think the release, the way that it's being released is more important at the moment. Obviously, there are implications here, right? If companies or countries cannot rely on their ability to access frontier AI from US foundational model companies, they will rush toward open source. They will rush towards other solutions.
8:06Big Technology Podcast Host:They may be slower to go to these more expensive frontier models, which means the ROI and the incentives to build them will be less. And therefore, we're now at a point where I won't say the entire business model is in question, but given the stakes and given the money that's being raised, maybe it is.
8:29Ranjan Roy:Well, hold on. Let's separate out. I think there's like a few different parts of that statement. The first part is, I guess, like on the or actually on the business model question. Let's remember that it was Anthropic that raised the flag and the concerns around this and basically invited the U.S. government to bring this up. Now it feels like this is becoming the default American rollout regulatory model for any kind of frontier model. but it was the lab itself that pushed the narrative open ai obviously appears that it's following it but do you think not a releasing a frontier model to like a full-scale audience is necessarily a bad thing do you think like the most qualified responsible people and companies should be the ones that start testing it and pushing its limits first yeah well we've gone
9:31Big Technology Podcast Host:back and forth in this i mean i i have in the past said okay maybe this is the right strategy to if you think that this thing can be this powerful then maybe you should be careful with your release uh so i understand i've said that here um but the other side of it is that like you you know you could end up in a situation where the government and the companies really pick the winners and losers in this space and from a business standpoint maybe they're able to charge a lot more right so the initial preview of mythos was quite expensive and actually the preview of gpt 5.6 is about half the cost of the of fable so that's interesting but yeah i think that like from a business standpoint um there are second order effects here and i think more broadly there's concern to me that if we're going to go into this moment where there is this you know government picking winners and losers situation that you could end up, yeah, you could end up with a more controlled economy, right?
10:35Big Technology Podcast Host:Like there have been people on Twitter talking about how like the US is much more top down on AI than even China is today. Here's another tweet that I thought was interesting. Someone wrote, I'm afraid we've entered a dark era in AI model development and access. And somebody else quote tweeted them and wrote, breaking member of the permanent underclass left shocked after finding out the permanent upper class is invite only. So I imagine that this moment is going to pass, but certainly there's like some worrying precedence here if it continues. Your thoughts?
11:07Ranjan Roy:Well, I guess I've found it a bit, the whole phrase winners and losers and the government picking them, because that is how the administration has essentially worked for the I mean, for especially in this second term. And it's just coming down on, I think, a faction that has kind of like screamed for libertarianism for a long time. But even though like when, you know, when it's who gets to buy TikTok or what any other decision or how does Stargate roll out or, you know, like all of these things, the government and the administration have been pushing. So it's just kind of taking the same approach to Frontier Labs in my mind.
11:49Ranjan Roy:And I don't know, I guess I have not been surprised at how things have been playing out at all. I don't think it's a good thing. I do agree that overall, like, especially like vis-a-vis China and how things are playing out overall across the entire global AI landscape. I think it's a very bad thing. But I guess I'm not as surprised as a lot of VC Twitter seems to be.
12:15Big Technology Podcast Host:Yeah, I wouldn't say surprised is the reaction I'm having. But, you know, it's important. I think it's important to talk about the problems that could lead down the line, both in terms of like, all right, if this is the new process, then certainly you're at a much greater disadvantage if you aren't able to use these models and your competitors are. Right. So who gets to decide that? I don't think that's the free market that the U.S. needs. Then the other side of it is, do you end up playing into the hands of your competitors? This is an interesting thought from Aaron Levy. And we've kind of touched on this, but it's worth talking about.
12:49Big Technology Podcast Host:If the U.S. remains at the frontier at all times and has heavy regulation on the release of intelligence, then we end up with an economic and geopolitical edge because we can control who has access to frontier intelligence. However, if we delay model releases and another player, specifically China, doesn't slow down and has equally strong models, not now but soon, then our delays end up advantaging their models and eventually their tech stack. And we've definitely seen conversations recently about how we're moving from people using a frontier model to using model routing, using more open source models.
13:27Big Technology Podcast Host:And I think this kind of goes to your point, right? If you think the harness is the big competitive advantage now and not the underlying model, then you could end up with this type of policy sort of icing out the frontier labs and having a lot of the benefits of this AI moment go to open source in China. Which, who knows, I mean, if there's a robust open source AI availability, then that could be really good in a lot of ways. But you also then sort of lose your ability to control on the safety side of things. And that's why you have people talking this week about banning open source models, which I don't think would be good either.
14:07Ranjan Roy:Well, I think let's dig into that a little bit. To me, they are two separate issues because especially a lot of the stories you're hearing about or work that's being done or even what I've been seeing with customers on the front lines around exploring open source models. A lot of that is more around cost and like what models are actually suited to what tasks versus the idea that anyone is going to use Mythos or Fable for a lot of the work that's being routed now towards open source. So I think like the cost driver, which we'll talk more about, is more of a factor rather than the regulatory side.
14:50Ranjan Roy:But I do think like, because we have to remember like the most state of the art cutting edge frontier model, that's really about whether it's, you know, finding tons of security loopholes or like pushing the boundaries of science, scientific innovation. Like, I don't know, like that kind of work is still to me very different than what we're hearing about on the open source side.
15:19Big Technology Podcast Host:right but i mean that's because the open source models have trailed in capabilities but it could end up i mean this might have happened whether we see the regulation or not right that the open source models will not only be cheaper but they can start to rival it's happening they can start to rival in capabilities and and that that really i mean especially now as these companies head towards ipo it really makes you wonder and i did try to speak with Brockman about this a little bit when he was at the AI summit, but like, it makes you wonder, like, if you have a model that's closed and is the equivalent of like 15 PhDs and has good EQ and can do stuff, get stuff done for you, and you have an open source model, right, let's say a couple years down the line, that's like the equivalent of 12 PhDs and has decent EQ and will still get things done from you, then like, what is the competitive advantage for the frontier models versus the open source if they all get that smart?
16:19Ranjan Roy:Well, okay. So on the business model and IPO question, and we're going to talk more about OpenAI's reporting on their potential IPO, but I think it's a massive deal. I think it's like an absolutely ridiculously massive deal. the fact that it completely destroys the entire story. The story has been expensive frontier models. All the ARR growth that was shown to the world from, call it September-ish, October-ish to February, March, was people just cranking on whatever the latest, most expensive frontier model was. The moment, and again, we can Jevons paradox it all day long, And I do firmly believe overall utilization of models and agentic AI is going to be just, you know, like exponential.
17:13Ranjan Roy:But the core IPO story, I think, takes a huge hit from this. So I think like the advantage of having something far superior for a short amount of time starts to go away anyways the moment people start realizing they don't need the Ferrari, the Honda Accord is great. Right.
17:39Big Technology Podcast Host:And, you know, up until this point, it was like a nobody gets fired for buying IBM situation where like you wouldn't look elsewhere because you didn't need to. because it was just kind of easier to plug and play these frontier models from the closed companies. But what I'm saying is because the frontier is now being restricted, it adds an incentive to look elsewhere. And when it adds that incentive to look elsewhere, you might see a slowing. Now, the foundational model companies would argue that their models are still better, which they are, that they have more compute, which they do. But ultimately, the expectations have been built up so high and the money is so you know significant in that they they basically need to shoot a hole in one here for everything to work or maybe a hole in two i don't know they have to basically be near flawless in their execution uh and the pressure is is now on and that's why i think you saw opening i'm taking the risky stance of saying we're going to play with the way the government wants us to play right now but we would really prefer not and you're right
18:42Ranjan Roy:Like they have executed dearly flawlessly. I mean, it feels like for the last six to 12 months. I mean, that's why the expectations have been built so massively. But yeah, I don't know. Overall, I think this is a big issue. I guess one thing I've been wondering, though, is right now you keep hearing and reading about as these frontier models actually, you know, like have widespread adoption, that's where the distilling starts to take place. That's where competitors can start to, I mean, like, you know, I've heard about these like distillation swarms where they're literally like asking, you know, like sending millions of prompts to try to like start understanding all of the model behavior and starting to work to actually recreate it from a open source standpoint.
19:32Ranjan Roy:Is there some world where the regulation is good to kind of protect the frontier model and you only give it to your hundred most profitable large enterprise and government customers that pay you the most money and also minimize risk of distillation as well could that could they be almost a little happy that that's the case my hot take is that one of the things that we saw from
20:01Big Technology Podcast Host:Anthropic with all these safeguards that it put on Fable was not as much that they were worried that Fable would be misused, although I'm sure there was some worry there. It was more that like, we know our competitors are going to try to distill this model and we're just not going to let them do it in places that provide the most value. And in fact, I think Anthropic was, you know, effectively coming out and saying that if we think that you're using our technology to build a competing model we're going to block your access to do that and they kind of backed off that a little bit because people were like anthropic doing any you know preventing any ai research with its models is anti-competitive and bad uh but yeah i think that's a that was a real worry and and this limited release strategy could prevent distillation to a degree although you know the models it's it seems like it's kind of hard to keep the models under wraps don't you
20:54Ranjan Roy:My favorite Bill Gurley this week had tweeted, if you're on the verge of AGI or ASI, why isn't your model smart enough to recognize espionage distillation in real time? You say you can cure cancer in a few years. Isn't sniffing illicit distillation quite a bit easier than curing cancer? And I kind of loved it because it's like, why do you need why? It is surprising to me that that is still such a threat when it is existential to their business that they haven't kind of invested or figured out a way to protect against this short of not giving as many people access.
21:35Big Technology Podcast Host:Yeah, no, it's a great point. I mean, I think that's what they were trying to do with these proactive safeguards. But obviously, it's still a problem. They just accused Alibaba. Anthropic just accused Alibaba of distilling their models recently. Okay, one thing on GPT 5.6. Then we'll move on to the OpenAI IPO stories. So they ran GPT 5.6 through a test called ExploitBench. And the capabilities were basically on par with Mythos. This is obviously a cybersecurity test. But Mythos is much less token efficient. So Mythos used something like 300 ,000 tokens for the test, and GPT 5.6 used around 100 ,000.
22:17Big Technology Podcast Host:So I would say it's at least two times more efficient than Mythos Preview, which is interesting, and it's priced more cheaply than at about half price.
22:30Ranjan Roy:Yeah, I mean, you see price constantly being now a highlight rather than kind of some afterthought. I don't think most people could have, you know, like reasonably known what their input and output token costs were in the past. And now everyone is looking, I mean, is looking at it. Wait, hold on. It's not publicly available in any capacity now, right? 5.6? Yeah.
22:58Big Technology Podcast Host:It's not publicly available. It's the same thing. There's a limited number of companies that have access to it.
Read the full transcript
23:03Ranjan Roy:But even for all three of them?
23:05Big Technology Podcast Host:all three like and terra all three okay so they're not doing like uh luna everybody gets a
23:11Ranjan Roy:little bit of access too but uh so you gotta wait yeah i mean i think the hope is that it comes out fairly soon that's just the way that they're talking but we'll see well i guess why why i'm asking that is in my mind and maybe it's wrong i still kind of associate the more expensive and heavy the model is the more dangerous it is and that's why when luna is fast and affordable for everyday work and that's kind of the positioning you would think like that should be easy and you just give it to everybody versus soul is the flagship model and uh that's the big dangerous one that can get us all into trouble but maybe that's the wrong assumption but yeah you got to
23:53Big Technology Podcast Host:be careful those nimble fighters right you think they're just lightweight and chill and next thing
23:58Ranjan Roy:you know they cut you they're they're they're exploiting software left and right bringing down
24:03Big Technology Podcast Host:global financial infrastructure exactly okay so speaking of global financial infrastructure we do have ipos on the way uh this is from but but the question is when this is from the new york times openai leans toward waiting a next until next year for its ipo openai is leading towards holding off its initial public offering until next year a turnabout that punctuates the uncertain future for fast-rising artificial intelligence giants. The maker of Chachipiti hired bankers and lawyers with an eye towards a public offering as soon as the third or fourth quarter of this year. But because that's not going to hit the$1 trillion valuation that Sam Altman has been seeking, it looks like they are going to potentially postpone.
24:46Big Technology Podcast Host:It's from the time story. Top of mind is what happened to Elon Musk's SpaceX after its IPO this month. It had the largest IPO ever, raising more than$85 billion at a$1.77 trillion valuation. However, it's fallen back to earth. I'll just share my conspiracy theory on this one and then turn it to you in terms of your views of the implications here. I think OpenAI knows it's important to beat Anthropik to the public markets given the growth that Anthropik has shown of late. it probably sees a better opening now because anthropic has been in a way held back by the government because of the fable fumble and open ai might be just sending out signals that it's going to wait but then it will pounce as a lightweight fighter does and cut cut anthropic get it get it get there first that's my conspiracy theory talking about two potentially trillion
25:45Ranjan Roy:dollar companies as lightweight fighters here but gotta be nimble gotta be nimble okay i i do like that because it was surprising to me that that could even be leaked and i feel a lot of this ipo reporting is very purposefully leaked because again in the past like and i mean having seen a lot of this very up close like everyone does whatever they can to not allow for any of this kind of reporting and we've certainly seen plenty of it. So it feels very like anything that comes out is purposeful. So then you have the question, why would they signal that they may be delaying? I do like your take there.
26:29Ranjan Roy:I think that's a fun one. But I also just think like right now, to me, what's interesting with the market is two months ago, everything felt like it had direction. Everything, the ARR was just unreal, like any kind of number that was being reported out. Overall, you know, everyone's feeling good about things. The Frontier Lab and state-of-the-art models still are the kind of dominant narrative. Meanwhile, to me, the biggest vibe shift has been when you see Bill Gurley or the Coinbase CEO, all brian armstrong like everyone is just bragging about not using the frontier models think about two months ago three months ago it was just oh my god have you used opus 4.8 or whatever like it was just gpt55 is now ahead of 4.8 that's all anyone talked about no one had ever even i mean most the average user had never even heard of quinn or like now with glm 5.2 this week So to me, it almost feels like this is a moment to wait because the entire market has shifted from two to three months ago and no one got out other than Elon.
27:49Ranjan Roy:And he did a great job at that. But I think waiting actually would be the smarter decision.
27:56Big Technology Podcast Host:But how long do you want to wait? Because if you wait too much longer, you might be in the situation where you've spent a lot of money. you're waiting for more money uh and your customers are starting to try to find ways to efficiency max their but they are they already are they already are but do you want do you want that you know right now the cut the curve is looking up into the right but do you want it to even out as these customers try to uh try to find more efficient ways to use your tools
28:29Ranjan Roy:I guess, I mean, and I have no direct knowledge either way on this. To me, the curve is no longer up and to the right from where it was a few months ago. Just everything. Yeah. I mean, just again, when the Coinbase CEO is bragging about model routing and here's five different models suited to different tasks and everyone is talking about that, it changes the entire growth curve. The entire story from September to February, March, April-ish, all of that has changed. It's just so clear. So to go out, you're going to have to show your latest results. You're going to have to show the last few months.
29:12Ranjan Roy:So I do think versus you have a lot of promising different business lines, start allowing a little bit more maturity and then go out with a stronger story rather than here's an ARR curve that looks insane. Right.
29:29Big Technology Podcast Host:And it's not just the Coinbase CEO and it's not just Twitter voices who are talking about finding more efficient uses of their models and looking for savings in their engagements with Anthropic and OpenAI. And by the way, there's also some reporting about Anthropic and OpenAI potentially overcharging customers. So it is a broader pullback, not just a micro-influencer pullback. and we have the reporting and we're going to talk about it right after this. And we're back here on Big Technology Podcast with Ranjan Roy of Margins. All right, Ranjan, so before the break, we talked a little bit about how there's this, now it's become a meme that people are trying to pull back on their costs with OpenAI and Anthropic and finding ways to do it.
30:17Big Technology Podcast Host:But the reporting shows that it's more than a meme, that it's a reality that seems like it might be more widespread than you would even know by seeing the tweets. This is from the information, how AI customers are lowing their Anthropic and OpenAI bills. As prices for Anthropic and OpenAI's flagship products soar, some large customers are using cheaper AI models from those companies' menus, as well as from other providers. Ensemble Health Partners, a provider of software for hospitals that plans to spend up to$100 million on AI this year, said it's had success switching to an OpenAI model that's one-twentieth as expensive as the company's more advanced models.
30:53Big Technology Podcast Host:The AI Power tool for writing appeal letters to insurance firms that won't reimburse Ensemble's hospital clients for care they provide. So the AI Power is a tool that does that. Ensemble sends about 15 ,000 such letters every month. And the chief technology officer of the company said switching to a lower cost model will result in savings of nearly$700 ,000 per year. Many Anthropik and OpenAI customers are willing to stomach the rising costs of these two companies' products, but some are hitting the brakes after blowing through their budgets. Okay, so basically what you're seeing now is it's almost as if these smaller flash models have gotten good enough that you can make these moves, and companies are looking at what they're spending and saying, hey, wait a second, this makes zero sense.
31:41Big Technology Podcast Host:We are actually just going to go to the smaller models and we're going to route and maybe use the more powerful models as a decision-maker, as a sort of top-layer decision-maker as opposed to a worker. Pretty interesting, Rajon, wouldn't you say? I mean, it does seem like this is, going back to our discussion in the first half here, a real, not to use a jargon, but a real economic headwind for these companies and not good, Again, if you're rooting for them to show exceptional growth as they get to IPO.
32:14Ranjan Roy:I mean, to me, the most interesting part of what you just said is you called it an economic headwind. But even within that reporting, they're still talking about using an open AI model. Just 120th is expensive. So to me, that's an even bigger challenge overall to the narrative is like you can be using Anthropic and OpenAI's own products and still that's not good for their overall IPO narrative and the business story because it's dependent on using the most expensive models. Not here's a whole suite of products and models and use all of them the best you can. And that's good for all of us. It really feels like the fact that that becomes a story in itself versus, oh, you're just using different parts of the OpenAI ecosystem.
33:06Ranjan Roy:I think that is a pretty telling thing.
33:09Big Technology Podcast Host:Is there a Jevons paradox side to all this where it's like, yeah, like you switched from using only our frontier model for everything to some of our cheaper models, but you're going to do a lot more now. And so therefore, a category that you never would have even spent on in 2023, now you're going to spend a ton more.
33:27Ranjan Roy:I think this is the crazy part of the timing of all this. To me, I firmly believe in a few years' time, and especially in the next decade, there is going to be just massive increases in overall token consumption, agentification of everything. I believe it. But the timing, because if this pressure wasn't so great on the two giant, the frontier lab giants, then you would think the product should actually very clearly, you ask it a question, you give it a problem, and it tells you. And I've seen Claude do this sometimes where it won't actively switch, but if you ask it which of your models are best suited for which part of this task, it will tell you.
34:16but they don't like core bake that into the product
34:20Ranjan Roy:because it's not good for the business overall and it should be they could own this and make their portfolio approach and everything is tightly integrated but it's given the IPO pressure that would be a dangerous thing to do over the next six months
34:37Big Technology Podcast Host:sorry wasn't the entire like big controversy around the GPT-5 rollout that it would route your queries to the appropriate model? So they're not doing that anymore? Or how does that comport with what you're saying now?
34:51Ranjan Roy:No, no, no. Hold on. Those are two separate things because the GPT-5 rollout, wait, wait, you're talking about when people were mad that it wasn't sycophantic and they wanted to go back to the forum? No, no.
35:01Big Technology Podcast Host:I'm talking about how when GPT-5 rolled out, this is just in ChatGPT, right? ChatGPT would have its own built-in model router and said, oh, this is an easier query. we'll send it to a smaller model or this is a tougher query we'll send it to our most powerful model and you really couldn't pick now you can pick i guess well no so okay i i would put those
35:21Ranjan Roy:in two different categories around building an agentic workflow that's repeatable and like predictable and they're spending the time to assign the right model to the right task makes a lot of sense i now okay now like when gpt5 did have their model router so yeah i i I acknowledge people were backlashing because they were like, oh, I'm getting dumber stuff. It's not doing this in the right way. When you're just having a back and forth, like turn based conversation, that's not where I think this is valuable. And actually, that's like a good example where it was clumsy the way it was actually executed.
36:01Ranjan Roy:And in that exact kind of like situation, it's probably not the most useful thing. I think it could be useful because, again, as we both talked about, there's times where you ask Claude a basic question and it creates a 20-page PowerPoint deck and just goes on, scrapes half the web and does whatever in God's name it's trying to do. and having a little bit of a lighter flash model typey thing might be good. But yeah, so to me, it's more as you have a workflow that is predictable and kind of like you know what the different pieces of work are, that everyone should be a lot more. And they should be trying to lead that and they're not because it's not good for them.
36:44Big Technology Podcast Host:Right. And by the way, as we're talking, again, this all matters for the IPO because the growth and the dollars are going to be pretty important. when we see these S1s and when they actually go out. Well, there might be some overcharges happening because of the opacity of the system. So this is from the information. Anthropic customers find errant charges, auditing startup says. Some customers may be paying more for Anthropic and OpenAI products than they should due to billing inaccuracies, according to Vaudit, which sells an AI bill auditing tool. Okay, well, obviously they're going to find this stuff.
37:19Big Technology Podcast Host:Okay, I'll just read it. We can talk about it. Between March and June, Vaudit audited bills sent to 60 companies totaling$34 million, mostly for usage of Anthropics Cloud Code, and found about$1.7 million in mistaken overcharges. What we are observing in that enterprise AI billing has become increasingly opaque, Vaudit said. Errant billing isn't unique to AI, of course, and can be seen throughout enterprise software cloud services and online advertising do you think that this is a big problem and ai is basically like you send these tools like cloud code out to do stuff for you and because it's so opaque you have no idea whether you're being overcharged or not i mean 1.7 million of overcharges out of 34 million is is not insignificant at all my favorite part of
38:10Ranjan Roy:this story is just hearing you say vaudit audited i love this name vaudit that was a tongue twister
38:15Big Technology Podcast Host:Better than Terra Luna, but not exactly the easiest thing to say on the air.
38:21Ranjan Roy:Vaudit audited bills. So I think this is an issue. I actually am not as concerned, I think, as normally I would think I would be around it. More just because like, yes, okay, out of 34 million, 1.7 is like a pretty significant percentage. But again, when no one has been checking their bills, it makes sense that without any malice, there would be errors happening. Like it's just if no one had been checking their bills, digging into their bills, even knowing what they're spending. When you hear Uber surprised that they blew through their entire year's AI budget in a quarter, that shows that no one was carefully tracking things.
39:09So, of course, there's going to be certain errors.
39:12Ranjan Roy:Actually, one of the ones that was the most interesting part of that to me, though, was when the session times out or you don't get an answer, which happens plenty, are you responsible for the tokens? because it actually kind of cited that as an error i hadn't actually thought of that as a billing error before it's just kind of something you you just eat like it's just right in ai you're not it'll time out sometimes and that's how it works but like that actually made me think are customers responsible for any tokens consumed during that session and that could start to be more and more of an issue yeah that could add up i feel like it happens to
39:57Big Technology Podcast Host:me all the time multiple like multiple times a day where i ask something to a chat bot and it thinks a bit and then gives up well whatever's behind that i don't know but it is it's a problem yeah all this stuff is going to add up as we start to see these companies head towards the most significant financial event of their lifetimes for sure and then of course we have the precedent this will be the last financial uh sorry financial market story we do um on on this on this topic, but SpaceX obviously not faring very well. I mean, it shed about, I think it was down 15 % or so on the week. It's trading at 153.
40:37Big Technology Podcast Host:That's the way it closed Friday. And all this sort of ebullience over the fact that it was on this legendary run has certainly quieted down. It went to two and a half trillion market cap. It was bigger than, I think, bigger than Amazon for a moment now it's dropped down to two trillion which is still nothing to sneeze at but definitely less frothy than its earlier trajectory seemed like it was going to be at this point your thoughts i two trillion is frothy i mean like that's why two two and a half it i will
41:16Ranjan Roy:say that first week of the post ipo there are moments in any kind of cycle you remember and And that's going to stick with me when you saw like, I mean, Elon Musk is a trillionaire now. He's heading to two trillion. It's bigger. And remember, this is a company with$18.7 billion of revenue total. I mean, Amazon, it surpassed its market cap. Amazon has$750 billion in revenue. Like this is a company that's losing$4 billion. So which again, if you believe there's going to be data centers in space and you believe like all that, you can come up with the story behind it. But when you just kept seeing it climb, it got to like 215, I think, 215 near the highs.
42:08Ranjan Roy:And then it's like surpassing Amazon, surpassing Meta, closing in on Microsoft. I don't know. those are the moments that whatever shakes out in the next year or so, I will definitely remember. Yep.
42:24Big Technology Podcast Host:Okay. Let's go to our last story, which is that Apple has raised the prices on the Max iPads by$200 or more on some models. Fascinating thing is happening right now. Memory prices have gone up and now Apple is making its product lineups much more expensive. This is from Bloomberg. Apple raised prices on its Macs and iPads early Thursday morning, a week after chief executive Tim Cook said the soaring cost of memory and storage chips would force the company's hand. The company briefly took down its Apple online store early this morning as it typically does when announcing new products. That was Friday.
43:00Big Technology Podcast Host:When it came back online, the price tags for Mac computers rose roughly 15 to 20 percent and iPad prices rose 15 to 25 percent. Among the price increases the base macbook air rose from rose 200 to 1299 the base macbook pro increased 300 to 1999 uh the entry level macbook neo increased 100 to 699 the ipad air increased 150 to 749 and the ipad pro increased 200 to 1199 dollars um i have become a somewhat of a lightning rod on X this week because this was my reaction to the news. I wrote that Micron's gross margin was 84.9 % last quarter and Apple's was 49.3%. Sure, price hikes are supply and demand at work, but at a certain point it's greed.
44:07Big Technology Podcast Host:I don't know, man. I mean, yes, the memory prices have gone up, but as somebody pointed out, if you're building a data center filled with these things, then it makes sense to raise prices. If you're building a device with a little bit of it, probably not. This is the exact tweet. This is from Carl Peterson. I sell GPUs for living systems. Integrators buying two terabytes of RAM for a server have an excuse for raising prices. Prices Apple buying 8 gigabytes for a MacBook does not. Greed or supply and demand? What's your take on the Apple price hikes?
44:43Ranjan Roy:Greed. Do you know what my tell was? I don't know if the HomePod mini, as an owner of multiple ones, has any kind of storage. But it certainly does not have any meaningful storage. So to raise that by 30%, I think shows that they're just going across the board here. I like, and also the increase in memory has been, if they're saying that this is here to stay, and maybe that's actually a positive signal for the SanDiscs and Microns of the world. But like, to me, you have this kind of massive price hike of an input, but to change these like long supply chain type products and pricing this dramatically, it's, this is actually pretty shocking to me.
45:36Ranjan Roy:You change the price of a flight because the cost of oil goes up. That's dynamic. That's relatively quick up and down fluctuations. This is huge relative. And once it's done, they're just going to keep it there.
45:54Big Technology Podcast Host:Well, unless they have a corresponding decrease in sales because they need those devices in people's hands for their service businesses to be strong.
46:01Ranjan Roy:Yeah, but we're all locked in and can't escape the Apple ecosystem. And then, oh, wait, maybe they are so confident that Siri is going to be so good that they're like, guess what? Everything's going up and you guys aren't going anywhere because Siri is going to become your default AI in life. That's what's happening.
46:24Big Technology Podcast Host:right i mean tim cook calling the price hikes unavoidable is just like ridiculous and you know obviously like the when the device goes up 20 that's the entire price of the device so dram could be going up by whatever it is but you're not you're not gonna it doesn't the increase in dram price does not account for a 20 increase in the entire device price it's crazy like they're going to make margin off that. And not only that, they're responsible for it to a degree. This is an interesting tweet that I read that sort of captures the situation. Apple spent a decade squeezing memory makers below costs.
47:05Big Technology Podcast Host:One of them survived, finally got pricing power, and Apple is calling it gouging. I mean, they didn't use the word gouging, but they basically implied it. Two years ago, Micron was selling DRAM below the cost of making it, gross margins underwater, and the kind of bust that has killed memory makers for 30 years. A field that once had a dozen players had dozens of players consoled down to three survivors. Micron is here because it executed the downturns better than the ones that died, not because anyone was generous to it. The buyer is now screaming about 84.9 % margins. Apple and the hyperscalers are exactly the players who spent that decade grinding memory to negative margins on every procurement cycle.
47:41Big Technology Podcast Host:Sorry, go ahead.
47:43Ranjan Roy:No, no, actually, I'm curious. So you, should Apple do this?
47:50Big Technology Podcast Host:No, I don't think so. I think it's bad business. I mean, first of all, it's customer hostile, right? You're using an excuse to jack prices up. Second of all, like it's going to hit your services business. You want services, the one thing that's really growing within Apple. I mean, recently we've seen some growth with iPhone. You're going to want a robust service business if you're Apple. And the way you do that is you get devices in people's hands.
48:14Ranjan Roy:But they've been actually, so they have been, I just had looked up the price of the premium iPhone model in the last seven years is up 65%. So it went from 969, which I think anyone who gets the new Pro Max, whatever, always has seen, it just keeps going up and up and up. And I mean, I think I am a example that I don't buy the new iPhones because I don't feel I need to, but I'm paying way too much in services businesses for things I forget about. So maybe in that case if you don't have the devices that's like a good example i guess it's just i don't know the confidence or maybe cockiness to raise prices this much and think it's not gonna just like crush demand they have to be modeling this out somehow or they have to something must be behind that
49:09Big Technology Podcast Host:if you're tim cook and you're trying to do john turnus a favor what you do is you maybe aggressively increase prices now of course the the products that are that are being jacked up they're they're not a massive part of apple's business but they're big enough
49:23Ranjan Roy:and the price will go up on the iphone it's ipad and macbook so if you're tim cook and you want to
49:30Big Technology Podcast Host:set john turnus up in the best place you jack prices up because of global memory shortage uh if the if demand remains somewhat consistent then you deliver turnus you know huge profits as he begins running the company if demand goes down you give turnus the enviable position of saying i am john turnus i am the price cutter i am going to cut prices to spark demand and you will love me now so either way option cook call option cook he's giving
50:04Ranjan Roy:yeah these giving uh turn is the greatest call option of all time okay i i like this one i like this uh direction it's a win-win um but yeah i i this i was genuinely shocked by this and trying to kind of like just chalk it up to uh chalk it up to short-term increases in memory prices that It might be here to stay, but still, I was pretty shocked. Yeah.
50:35Big Technology Podcast Host:I mean, someone told me, like, do you think it's a charity project or something about Apple when I said it was greed? And I said, obviously, get your bag if you're Apple, but it doesn't mean you're beyond reproach and I'm reproaching. I mean, of course, it's not a charity, but you don't have to, like, cheer on companies that are greedy. This is from Scott Heiferman, the founder of Meetup. Apple is, was a change the world project. Make insanely great products for as many people as possible. Good profit makes sense. Beyond that isn't the point of the project. Asking if it's a charity project or something is the question statement of the truly naive.
51:15Big Technology Podcast Host:Preach, Scott, preach.
51:17Ranjan Roy:I like your reproaching. I am reproaching. I am reproaching. I'm reproaching you, Tim.
51:23Big Technology Podcast Host:I'm reproaching. and uh i don't know if they don't let me into the next wwc i i really couldn't care less uh
51:30Ranjan Roy:too much reproaching no reproaching if you want entry but you will reproach you'll reproach
51:37Big Technology Podcast Host:i'll reproach all right two things before we leave um i want to say thank you to you ranjan and thanks to everybody who joined us at the summit last week we've obviously been putting some of the interviews up we'll have the greg brockman interview in full up this coming wednesday It was a good time, wasn't it?
51:55Ranjan Roy:I had a great time. Thank you for everyone who came out. And it was fun. We do this virtually, sometimes in person together, but just meeting all our listeners in person. That was a lot of fun.
52:09Big Technology Podcast Host:It was really cool. So thank you, everybody. And if you weren't there, obviously, we're trying to get as much of the content to you as possible. So we'll have the Brockman interview this Wednesday. and then we will go back to our normal podcast style, aka in studio or remote, one-on-one conversations with no audience starting the following Wednesday with Meta CTO Andrew Bosworth. So stay tuned for that. All right, round jump before we go. Tech industry lost a giant this week. Om Malik, the blogger, founder of GigaOM, VC at True Ventures, passed away on June 24th. This is from his blog. He passed away at Stanford Hospital after a long health journey with his heart.
52:55Big Technology Podcast Host:He was surrounded by friends and family. Om was 59. Just a word about Om. I mean, if you've seen any of the tributes online this week, you could tell that he was very generous with his time and his advice to anybody in the industry. you know i he's been on the show of course we had a great conversation with him not long ago and we definitely lost one of the good ones and uh the the night the nice thing about om is he really he seemed to care about everybody in the industry i mean when he started talking to me i was like wait om knows what big technology is and he knows who i am that's the craziest thing but he was read in and humble and um and kind of blunt and like uh in a fun way i remember I was sitting next to him at WWDC or some Apple event, and he had been writing these columns from The New Yorker, and another journalist was like, Om, how do you write for The New Yorker?
53:50Big Technology Podcast Host:And he just looked the guy in the face and smiled, and he goes, because I'm good. And he really was. I don't know if you have any memories or reflections on Om, but definitely he kicked off the modern tech blog era for sure.
54:06Ranjan Roy:No, so for me, having never been a full-time writer, but writing plenty, I'd gotten in touch with him a number of years ago. never met him in person but when i was start running a startup in 2013 and just you know like trying to get into tech crunch and giga ohm and stuff like that what amazed me is he responded and we actually exchanged emails and he was just like you know friendly versus so much of the i mean just overall ecosystem as a whole and i think the most important it's like yeah he he blogged I mean, he got me interested. He was one of the first blogs I started to read and actually got me interested in the whole new media thing.
54:55Ranjan Roy:So, yeah, seeing all the outpouring and everything, it was pretty special. Yeah.
55:02Big Technology Podcast Host:I don't know if this – I'm going to just read this. So Katie Jacobs-Stanton, who was a VC and was a Twitter executive, posted this letter that Om wrote to her when her father passed away. and I think that like when people experience loss sometimes it can be difficult to like to hit the right tone and I think this letter that Om wrote really did hit the right tone I'm just going to read it and we can sort of end with this he wrote dear Katie I read the sad news you shared last night on Twitter and I wanted to take a moment and send you my thoughts prayers and hugs to get you through this tough time we all have a habit of mourning what we don't have instead of celebrating all the gifts we are bestowed upon.
55:43Big Technology Podcast Host:His life was a gift to you. I didn't know your father, but it's clear that his grandness lives in you and yours. Your compassion and empathy for the human condition is a gift of your parenting, and I hope you can remember that when there is a moment of sadness that shakes up your world. I hope you find his memories and his lessons a guiding light for the future with much love and prayers for your family. I mean, just beautifully written and honestly a lot of the same can be said about Om today so rest in peace Om, one of the good ones I know we typically end on a laugh but I think that that yeah Om's presence in the tech industry, around the tech industry brought a lot of joy to a lot of people and so we'll do our best to live his legacy on in the ways that we can
56:39Big Technology Podcast Host:okay well we'll see you next time on big technology podcast
56:45Ranjan Roy:this episode is brought to you by google chrome you think you know a browser but gemini and chrome that's new it can help you with practically anything on the web like restoring a vintage motorcycle from a 50-page restoration block or finally break down that long article you've had open for weeks gemini and chrome is here for it ready to make anything online make sense There's no place like Chrome. Check responses set up required compatibility and availability varies 18+.
57:11Big Technology Podcast Host:This episode is brought to you by Google Health. Stop chasing someone else's definition of health. What matters is what's healthy for you. Google Health offers a new kind of coach,
57:20Ranjan Roy:built with Gemini for effortless tracking, sleep insights, and holistic coaching tailored to you. Visit googlestore.com to learn more
57:27Big Technology Podcast Host:and start a new relationship with your health. Requires Google account, Google Health app, internet, and Google Health premium subscription. Features subject to change. Availability and results vary. Not intended for medical purposes. Works independently of Gemini apps. Check responses for accuracy.
From the publisher
Ranjan Roy from Margins is back for our weekly discussion of the latest tech news. We cover: 1) Mythos is back 2) OpenAI releases GPT-5.6 to a small group 3) Should the government pick winners? 4) Is this a blessing to open source AI? 5) Is the gating of frontier models bad for these companies businesses? 6) Frontier lab customers are finding cheaper ways to do business 7) There's a bunch of overbilling happening in opaque AI systems 8) Is SpaceX's valuation frothy? 9) Apple's price raises: Greed or not? 10) Apple's impact on the memory space 11) Rest in peace, Om Malik
---
Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice.
Want a discount for Big Technology on Substack + Discord? Here’s 25% off for the first year: https://www.bigtechnology.com/subscribe?coupon=0843016b
Learn more about your ad choices. Visit megaphone.fm/adchoices


