Big Technology Podcast - Will Apple (Finally) Get AI Right At WWDC?, Anthropic’s Worry, Microsoft vs. OpenAI

Episode Date: June 5, 2026

Ranjan Roy from Margins is back for our weekly discussion of the latest tech news. We cover: 1) Does WWDC matter, or is it all about iPhones? 2) Apple's plan to build AI into the operating system 3) Apple finally seems to be building to the technology's capabilities 4) But... Apple may still waitlist people for forthcoming features 5) Anthropic says AI might be close to being able to improve itself 6) Is that marketing? 7) Can Anthropic stay ethical as a public company? 8) Microsoft and OpenAI are now in direct competition 9) Microsoft is building models, is OpenAI building a cloud? 10) Let's go Knicks Join us for the Big Technology AI Summit: https://summit.bigtechnology.com --- 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

Will Apple finally get AI right at WWDC? Anthropic is worried about runaway AI improvement, and Microsoft and Open AI once partners are in head-on competition. That's coming up on a big technology podcast Friday edition right after this. I'm just back from ServiceNow's Knowledge 2026 in Las Vegas, and the conversations I had there are ones you're going to want to hear. I sat down with their president and CPO Amit Zaveri on the platform strategy powering enterprise AI, chief people and AI enablement officer Jackie.

Caney and Chief Digital Information Officer Kelly Romack on what AI really means for the workforce. The technical leaders behind ServiceNow's NVIDIA partnership on shipping AI at scale and Alta Beauty on deploying ServiceNow's technology across 1,300 stores. If you want to know where Enterprise AI is actually headed, not the hype, but the real story, you can find these videos on my YouTube channel. Search Alex Cantorwitz on YouTube. Depending on who you ask, between 80 and 95% of Enterprise AI projects fail. To get AI to work for you, you don't need more tokens.

You need better people. A board pairs powerful proprietary tools with senior engineers who've seen it all. That combination means your project doesn't stall, doesn't drift, and doesn't fall. It ships. Whether you're a startup that needs to get to market or an enterprise with complex legacy challenges, a board delivers exactly what your business needs fast. Aboard is your partner for AI transformation. Visit abord.com and let's build something together.

Welcome to Big Technology Podcast Friday edition, where we break. down the news. In our traditional cool-headed and nuanced format, we have a lot of news to break down for you this week, including a full preview of seemingly everything that's going to happen at WWDC. Apple's big cornerstone event coming on Monday. We also have news that Anthropic is worried about runaway AI improvement. I'm sure Ranjan is going to buy that hoke line and sinker. And then we're also going to talk about Microsoft and Open AI once partners now competing head on. Joining us as always on Fridays, do it as Ron John Roy of margins. Ron John, good to see you.

All right, you already got me there, hook, line, and sinker. I was pasting that story into our document, and I was with every paragraph, like, anticipating your reaction, saying that it's marketing. So we'll have an interesting segment coming up. Our prep doc is just one long opportunity for you to trigger me, Alex. Starting with Siri. Starting with Siri today. Then go to Anthropic and then, I don't know, where you're used to.

stand on Microsoft opening eye. All right, but let's build some tension here. It's going to be an exciting final segment. Okay. Let's build the tension here. Legendary debate. Okay, but seriously, you should talk about the Siri thing. I'm actually, so, so I'm about to fly out to San Francisco for, well, I thought it was going to be WWDC. It looks like I am on the band list this year, but anyway. Wait, really? I'll be in proximity if anything else. Wait, wait, hold on. I'm trying to remember. Did you go last year where you already disinclated. I have been. Yes, I've been to WWDC at least the last two years. So this year, no luck. No? No luck. But I'm going to, I'm going to, it's not Tim anymore. It's John. Tarnas.

All right. I'm going to cross my fingers and maybe I'll get in and have a report for you from the ground then. But you know what? Maybe you don't even need to be in Cupertino to get the WWDC rundown because it looks like Mark German has like every detail of what Apple is about to announce. And we're going to run through all the big announcements that Apple is going to make at WWDC and give you our perspective on it. And I'll start off saying that I read this and left immediately extremely optimistic about where Apple is heading. It seems like the company is finally understanding the way that it can use artificial intelligence

in its products. And the fact that it doesn't need to build everything. And let me start with, before we get into the actual announcements, Ranjan, if you will, let me start just with the context here for Apple. Because, you know, while we've said the WWDC and the AI work that they're doing is the most important thing the company can do at this moment, at least in the near term, it hasn't been the case, right? Company is surging right now on the back of iPhone 17 sales.

It did 140 plus billion in the first quarter, which is sort of the traditional Q4 for everybody else, $111 billion in the most recent quarter. Its stock is up 56% over the past year, and this is all pre-AI. There's no AI competitor device. So before we get into this Siri announcement, because we will spend a lot of time on that, let's just talk about the big framing here. Am I right in thinking that we shouldn't get too crazily concerned for Apple speaking specifically about the company and the business about AI?

Because up until now it hasn't factored. No, I think Apple still is an iPhone selling machine, even though myself and plenty of others have questioned whether this is actually a sustainable business model. And I have not upgraded since the 15, even though I bought every new iPhone for like a 10-year period. I mean, it's clear the business is healthy, at least on a relative basis. The core businesses, I mean, when I say healthy, it's still just generated. ungodly amounts of revenue on a continued basis. So we have to give them that separate from all the entire AI conversation. And to their credit, they're doing it without any genuine innovation on the actual product

side either in a number of years. So Apple, congratulations on just having us all locked into your ecosystem and continuing to sell us products. My left AirPods keeps buzzing. And I'm probably just going to go out and buy the AirPods 3 Pro. So they got us. Do you think it's healthy? What?

The business? The business. Let's start there. Yeah, absolutely it is. And by the way, I think you're underrating the 17. I think the 17 is a crazy machine. And it's much better than any phone to come before it. So I'm happy to have it.

But okay. Hold on. Sorry. You are saying, do you believe there's been genuine product innovation from Apple? Are you going to stand by that statement that the iPhone 17 is truly a revolutionary product? No, I've always said they, they're, are refiners, not revolutionaries under Tim Cook, and that's been the case. But they have refined. They've made the camera better. They've made the battery life better. They're computing more powerful.

And through refining all this, they've made a, yeah, a great device. And by the way, the revolutionary devices, no one's built them, despite Open AI trying to do it, despite meta trying to do it, despite Google trying to do it, despite Amazon trying to do it. We don't have any revolutionary AI computing device yet. It's not here. I still, the meta, the meta, the Metaglasses, I genuinely, I consider, okay, it's not like, maybe it's not as magical as when the first iPhone came out, but there's devices out there. I'm going to put, let me think, meta glasses certainly. The Apple Watch, I will give credit, was in its time revolutionary as well. AirPods when they first came out.

Foldable phones right now. There are, it's happening just now with that. Let me reframe this for both of us, right? Okay. Okay. Maybe the revolutionary AI device is going to be the iPhone. And when you look through the announcements, you start to see a different strategy from Apple. The first was, you know, the Apple intelligence rollout two years ago was this big rollout, like, let us dream up what AI could do and sort of put together this visionary, ambitious plan.

This is a much more scaled back version of that. But it seems to be at least in accordance with what the technology can do. So let me give you a couple of examples of what's going to happen. This is from the German story. There's a new Siri. It was known as Project Campo internally. It sits at the center of Apple's renewed AI strategy. The idea is to transform the assistant from a voice control system to a do-it-all companion,

letting users handle tasks across iOS, iPodOS, and MacOS through the day. It will also be easier to control both in-house and outside apps using Siri. To power this new assistant, Apple is relying on a Gemini model from Google. the company that is both a rival and a long-time partner. In addition to using Google's underlying technology is part of a billion-dollar agreement, the company is hosting much of the new Siri on Google servers. Apple has long touted its ability to safeguard user data,

so the arrangement may spark privacy question. Okay, they are expanding Siri beyond this sort of weird magic trick to actually interface with devices, but now here's a thing. Let me say it was one last thing, and then turn it over to you. What I am impressed by with Apple is they are realizing that they have an interface advantage. They own the mobile operating system, at least for iPhone users, and they're going to use that to flex and say, all right, AI device.

You own an AI device? We'll build you an AI device. We'll build it right within our operating system from the German story. Apple plans to let users swipe down from the top center of the iPhone to launch a new search or ask interface. The new command will open a revamped experience designed for getting things done or search. searching by typing through voice, though voice control remains an option in the search or ask page, users can launch apps, start text messages, ask about the weather, add calendar appointments, sift through notes, trigger shortcuts within apps, or search the web using AI.

That to me seems the right way to do it. I think, all right, let me try to be not political here, but let me try to be nuanced. This is the most basic AI stuff imaginable. This is like the most basic user experience stuff imaginable. Like, yes, we are all able to dictate to our phones. I know Apple's native dictation is bad, and I use whisper flow. I do most of my search via AI as well. Again, but I do recognize, as you said, the iPhone could be the AI kind of like interaction layer.

And there's still a lot of people who don't. use AI, like really in an integrated manner in their day, and Apple still has that opportunity. But not, and maybe it's good. None of this feels, all of this stuff feels like stuff that is so basic to me that it's kind of shocking that this feels like an announcement, but maybe that's actually good because that they can actually get it done rather than just having Bella Ramsey on an ad two years early. Okay. So I know it sounds underwhelming. And, you know, as the guy who just got disinvited from WWDC, you know, I'm far, not the first one to take Apple's side on this one. But let me explain to you why I think that this makes sense.

Okay. Alexander Wang, the guy who's running meta-super intelligence labs, was at the Bloomberg tech conference this week, whatever it was. He said, this is a quote from him, ultimately what we're really excited to build for the world are the best personal agents. Right? The thing is, to get to Meta's agent, you have to open up your iPhone and tap on some agent functionality. We've talked in this conversation already about a swipe down in the operating system being something that will immediately take you to Apple's AI experience where you can search and ask. That's going to be powered. By the way, they're going to bring in lots of different models. So there's going to be a model switcher that will enable you to pick the model you want.

And at the top, I talked about how, what is Siri going to be, a do-it-all companion, letting users handle tasks across iOS and control in-house and outside apps using Siri. This is the vision of the personal agent lived out within iOS. Even if meta's is better, even if OpenAIs is better, they're not going to have access to the operating system. And that gives Apple a chance, building on top of Google technology, which isn't shabby. to sort of use the force of its advantage and maybe, if not win, at least factor heavily into this AI race. All right. I'll give you that. I mean, just the value of inertia, the value of being where customers already and users already are. There is tremendous value, especially at like a mass scale in the general population with AI. I think there is a huge opportunity there. And actually, I'll admit, I have started using Google.

Google's follow-on AI mode where you do a Google search and then you ask questions after. And I almost feel like dirty doing it. It feels like the most like kind of just hacky, clunky. It's just not a good-looking interface. But I actually have started using it more just because I still will every now and then go into Google search to quickly search something. So the value of being there, I think, is good. Where I would still caution. and actually back to that example around meta and personal agents,

where Apple, I think, is potentially getting into trouble is even, and I honestly hated this paragraph where it's like customers will be able to do more advanced tasks like telling Siri to write an email by giving it topics and information and asking it to pull the message together. Like, come on, that that is every, that's GPT 3.5 right there. But where it's... We're not picked into the operating system. No, but here's where it's going to get technologically challenging.

Having spent the last year and a half building agentic systems at Ryder, I know where things can break in where things are more stable and easier. The idea that you're going to be able to access any app on your phone and data across all these various systems is actually a very technologically challenging thing. I think that's probably what's held Apple back so far. Like, if they just limited it to, like, it can't even be email because then does Gmail, Gmail owns your email side of it. They keep talking about, they're like, for the first time users can ask Siri when they're available for appointments before scheduling something.

Does that have to make sure mean that your ICAL is connected to your Google Calendar if you're a Google Calendar customer? Just managing that data, knowing which app to call related to a specific. task is a lot easier when it's a small universe of options. If they're really going to kind of own those kind of data routing systems across your entire phone, that's actually a hard technological problem to solve and can get, adds a lot of unpredictability. So do you think they're going to pull it off? Or how do you think they roll it out? Do you think they do it? We're just going to have email and calendar appointments on day one. Do you think I can pull up a YouTube video via prompt, via voice prompt.

How do you think they're going to do it? Okay, so those are two separate questions, and I think they're both worth answering separately. So, you know, the question is whether they can, the first question is whether they can pull this off. And, you know, it's interesting because I was reading through the announcement. Here's another one that they have in, well, it's not the announcement. It's just German, but Gervin's that good.

Here's another one. The company is planning to allow users to throw multiple. commands at Syria at once. The feature will let people combine requests. For example, asking Siri to check the weather, creating a calendar appointment, and send a message all within a single prompt. And I read that and I was like, this type of things that they're talking about, they're catching up to where the technology is, right? When you're in chat GPT, it does have an understanding or in Claude. It does have an understanding of like it can handle multiple prompts at once. It kind of knows where to go.

if Apple can mirror that, then that makes a lot of sense. The question is, will they be able to? So before we get to how they're rolling it out, I think you're really keen to hit on this because or really, really incisive to hit on this. Because ultimately, it seems like the ambition is not out of step with where the technology is. The question is, can they execute? Do you think they can? I, what do you think, first? The only answer I have is they haven't shown us they can. Okay, I think that was a good answer. That was a good one.

But the vision is not ridiculous. I agree. It's not ridiculous. It's not ridiculous. Any company within reason of their scale and history should execute on this. But I like how you put that. They haven't shown us yet that they can. So then what's your gut check in terms of whether they can or they can't?

You're the one that's been asking for better Siri for, forever. I want them to. That's why it's like, I don't want to say an abusive relationship, but it's been so many times I've wanted to believe in them. So many times, starting with Bella Ramsey kind of like ignoring, not listening to her dinner companion, but still, I think, writing an email to them or whatever that ad was two years, two years ago. I mean, the fact that they actually position themselves as being able to, to provide this as a service two years ago. I think actually now, now it's even make me think how fundamentally they misunderstood the actual technology back then.

I just hope, I want them to, I don't think they can. Right. So the question is, so this is progress for Apple. This is what I'm kind of saying. They have made progress. They understand the technology. They're actually showing us a vision that aligns with the technology's capabilities. And they probably have Google, whether it's, it's forward deployed engineers or whatever you want to call them or Demis himself, they sure have Google in there working with them to ensure that this is deployed successfully because it's Google's reputation on the line as well. Google doesn't want to be part of a failure here.

So the execution part is a big question, but at least this is something that's a target that is reachable. Actually, how do you think that Google Apple relationship works? And the reason I ask is I have a pixel nine. It's not a burner phone. I just got it at some point, I think, relatively inexpensively, and I wanted to, like, have an Android phone as well for testing. Like, the way Google Assistant is integrated into the overall phone is all of this. It's done all of this for the last year, and it does it well.

Even Alexa Plus on the Echo Show does all of this really well. So it should be table stakes. It's where the technology is today, but is Google incentivized to me? make this work? Well, I don't even, not even looking at the agreement you would imagine they would be, because it's Gemini's name. It's going to be there. Well, and it's consumption at a massive scale.

Like, actually, is Apple going to pay Google on a consumption basis? The more people use it. No, no. The, sorry, the stories seem to indicate that they basically distilled their version of Siri from Gemini. Okay. Well, but then the better this gets, the worse it is for Google's hardware business. Like, this still represents, again, I had not thought about moving from Apple and I get so frustrated with Siri that I've looked up the pixel fold a number of times and I'm not ready to drop 1,800 bucks and completely remove myself of the Apple ecosystem.

But that is a threat to them. And that is an opening for Google. but I guess they're basically, they just signed away the model kind of adaptation rights to Apple. But beyond that, they don't really care is what we're saying. Yeah. I mean, ultimately, if you're Google, right, and this becomes, are you asking if this becomes a defining feature of the phone, is Google hurting themselves by selling this to Apple? Yeah.

Yeah, yeah, yeah. Yes and no. I mean, yes, they're going to power like they're there. There are rivals, right? It's like number one and number two in terms. of phone makers, but they also have great power over Apple, right? Because they're going to make another more powerful model and Apple's going to want to distill that and they're going to make another more powerful model. No, that's true. And Apple's going to want to distill that. And at a

certain point, if they see that, oh, we have a chance to like decimate the iPhone, I don't think they're going to be making all these deals. So it's great power in the hands of Google. I like that. I like that. Okay. So now that I've made the case that Apple is thinking the right way and has a good chance of being able to execute this, even though we haven't seen them do it before, so we're going to, you know, hold our praise until we see it. There is a disconcerting line in German's report that I'm going to read to you because you asked about the rollout and then get your reaction. Apple has labeled the new Siri as a beta and preview internally, suggesting that the assistant won't be marketed as a fully finished software when it's released later this year. The original

Siri held the same description for two years. There is also the possibility of a wait list or some sort of people who want to, of some sort of people of some sort for people who want to try new features and approach, an approach used with the initial launch of the Apple intelligence platform in 2024. Your reaction. Oh, God. I was just getting excited. And this, okay, hold on.

The upside, I would prefer they actually, more strongly take this approach rather than having a heavy glossy marketing campaign that indicates that everything is perfect. But this is just giving me PTSD of that first Apple intelligence launch. Again, I think nearly two years ago, I remember sitting there and like enabling Apple intelligence on my phone thinking it was going to do something. And it really is one of the worst kind of product rollouts. I don't want to say in history, but it's been pretty bad. So I was just starting to get excited that maybe they had a really coherent strategy and we're believing in the execution. But I'm going to wait and see. I'm going to wait and see

hope. But that just made me a little more uncertain. Do you think, does this make you more or less excited? Less excited. I mean, I definitely held that out to caveat at the end. I would say if I were to sum up this my perspective on this it is three things one AI doesn't matter as much to Apple as you know the narrative makes it seems like at least in the short term we're really going to have to wait for an AI device to really matter for Apple to start to say that their failures on AI are going to really stack up and add to anything now that's not that doesn't mean they're going to be safe forever but they're at least safe for now so that's point one point two is is I think that the positive sign, like I said before, is that they are actually building

to the technology's capabilities as opposed to like something they dreamed up, you know, in a conference room in Cooper Tino together. But then there's the worry. And the worry is the execution. And while they should be in a better position to execute this time, that line about the wait list worries me. Worrys me for sure. Do you think the, failure of execution has been, what's the term, is it Dutch disease where like, it's where you have oil or, oh man. Natural resource curse? Okay, it is Dutch disease. And originally it was around the massive influx of foreign currency.

But yeah, the idea, it's like a resource curse that when you have some specific resource, it kind of distorts your ability to kind of develop other parts of the economy. within a nation. It shouldn't be because, oh, go ahead. Go ahead. Well, no, I mean, those iPhones are still selling. They haven't, I mean, maybe you believe they have dramatically approved. I would tend to argue that it's been, yeah, at least you said refining, not revolutionizing. But again, when you're selling iPhones at that clip and everyone keeps buying them and I'm about to go spend 250 bucks to replace an AirPods.

like you just don't have the internal energy and incentive to actually build these kind of things. If you look throughout big tech, you will see companies with natural resources. And the ones that these tech giants have been able to stay on top because they haven't sat on their laurels. And they've realized they have a natural resource, but they still need to innovate. That goes to everyone like Google, for instance, with search. greatest natural monopoly of all times. Exactly. And you can see that they're, I mean, they've been big on AI for the last decade. So to me, it's not, you know, maybe it's part natural resource curse for Apple, but I shouldn't,

I don't think we should wave it away as saying it's just a natural resource curse. I think it is a cultural issue, and I've been on record about this for a long time. The culture of Apple is very high on its own supply. And they believe that the culture that they've built is responsible. for the creation of the iPhone and while it was that also came from Steve Jobs and so they've remained this refined jobs idea culture even though they don't have jobs anymore and they are sort of so happy about the way that they do things or proud of the way they do things that they will not try any other model that's why yeah you're I mean yeah Google

completely disproves my point. But I still, one of my favorite hot takes is that Google's entire AI transformation success can be traced back to the effect that Sundar was a McKinsey alum and looks at everything in terms of like organizational design and change management.

So basically if Tim Cook was McKinsey, they'd be fine. I stand by it. This is a pro-McKinsey podcast now. It is interesting because, The McKinsey model of like we talked about the engine room and farming it out to the product areas, I just spoke about this with MG on Monday. It might help make the specific products better,

but the question is, can you build a unified assistant with that model? Because you still have this centralized division working with the product areas within Google, like Gmail and maps and search. But if we're going to move towards a quote unquote super app, you know, you're going to need these product areas to subcede their egos to sort of make that centralized app work. Right. Again, like you were talking previously, what's the cutting edge now? It's learning what to call, right?

It's learning what systems to tap into. And if you have a lot of systems that, you know, sort of prefer that people spend time within them or prefer priority over others, it's going to be tough in the centralized assistant world. On the topic of super app, in a bit we're going to get to, uh, one of the other stories where we'll discuss it more, and I have some gripes. But one of the most interesting parts of this was one of the biggest additions this year is a dedicated Siri app for holding conversations and continuing past chats. Users can jump into the Siri app by pulling down on a result.

So basically dedicated Siri app, running conversations, basically your chat GPT-ish style assistant, And do you think Siri is going to become the super app one to two years from now? I think that's Apple's ambition. I mean, that's part of the ambition for sure. Do you think there was a slide deck Siri, the super app? Super app. Somewhere.

Super app Siri. Probably. I mean, the truth is iOS is a super app, right? So these apps become operating systems of their own, right? That to me is like the real question in terms of the viability of Apple's business model moving forward. It's if the chatbot and the AI becomes the operating system, what's your answer to it? So you would imagine that they're thinking about that exactly. Super abscary.

Super abscary. Okay. So we'll obviously watch and come back next week and talk a little bit about what we've seen. In the meantime, we should definitely talk about anthropic telling us that AI may be close to recursive self-improvement and maybe it's time to slow it all down. And then, of course, Microsoft and Open AI, once partners. now starting to battle it out for the crown. That's coming up right after this. Summer always changes how I get dressed.

I want pieces that feel lighter and more comfortable, but still put together. That's where Kintz comes in. They focus on well-made essentials and breathable linen and soft organic cotton. The kind of basics you will keep coming back to. Everything at Kints is priced 50 to 80% less than similar brands because they work directly with ethical factories, cutting out the middlemen, so you're paying for quality, not brand markup.

Personally, while I'm moving between air conditioning and summer heat, a lightweight zip-down sweatshirt from Kintz has been a life-changer. Elevate your summer wardrobe. Go to kins.com slash big tech for free shipping on your order and 365-day returns. Now available in Canada, too. That's Q-U-I-N-C-E.com slash big tech for free shipping and 365-day returns. Kintz.com slash big tech. Most leaders know how work is supposed to happen,

but when it comes to how it actually gets done day to day across tools, teams, and handoffs, they're mostly guessing. That's exactly the problem Scribe Optimize was built to solve. Trusted by over 80,000 enterprises, including nearly half of the Fortune 500, it gives leaders a live view into how work is really happening across approved business apps without interviews, manual process mapping, or extra effort from the team. And because it's continuously analyzing real workflow activity, the insights stay current instead of going stale the moment a process changes.

You can see which workflows are happening, where time is going, and which tools are involved. It automatically surfaces top issues, explains why they're happening, and even recommends ways to fix them with estimated time savings. And importantly, it's built with privacy in mind. So activities only captured in admin-approved business apps, and user-level data is anonymized by default. The kind of visibility that used to take months, now it's just always on. If you're ready to stop guessing and start seeing, visit scribe.

That's s-c-r-I-B-E-D-H-R-I-B-E-D-H-R-I-E-D-E-S-B-T-E. Insurance isn't one-size-fits-all, and shopping for it shouldn't feel like squeezing into something that just doesn't fit. That's why drivers have enjoyed progressives name-your-price tool for years. With the name-your-price tool, you tell them what you want to pay, and they show you options that fit your budget. Enough hunting for discounts, trying to calculate rates, and tinkering with coverages. Maybe you're picking out your very first policy. Or maybe you're just looking for something that works better for you and your family.

Either way, they make it simple to see your options. No guesswork, no surprises. Ready to see how easy and fun shopping for car insurance can be? Visit progressive.com and give the name your price tool a try. Take the stress out of shopping and find coverage that fits your life on your terms. Progressive Casualty Insurance Company and affiliates. Price and coverage match limited by state law. And we're back here on Big Technology Podcast Friday edition with Ron John Roy of margins, as we do every Friday.

Ranjan, I thought of you when I saw the headline of Anthropics' latest blog post, or really an article, the headline is, when AI builds itself our progress towards recursive self-improvement and its implication. This is from the story, Anthropic engineers on average ship eight times as much code per quarter as they did from 2021 to 2025. that is, of course, with the assistant of AI models like Mythos, the rare occurrences of misalignment present in today's models could compound as the models build their successors, growing more frequent, but less understood until we lose control of them. The argument is basically AI has made engineers much more productive than they were previously.

They're still a human in the loop. There may not need to be, and if there doesn't need to be, there could be some problems. What did you think when you saw this and the numbers, the post, the ideas? What was your reaction? I can't even right now. I actually, I think one of the things that it is funny, I feel we're at the point of the market cycle where I'm like going to be citing Ed Zittron more, who has been quite a bear. And I still vehemently disagree with him on the potential of the technology itself. But I liked he had said after this announcement, it's like, get some fucking new.

material, please. And like, I mean, it's just, like, we have heard this over and over again. I talked a lot about this a few weeks ago when, like, the mythos announcement was so, and a developer eating a sandwich in the park and receiving a message that had broken out of containment was a very clear kind of PR-coordinated message push. It's my question. too anthropic is as they are steamrolling ahead and obviously like to me the one of the most interesting stories of the next two to three months is who gets out first and when space ex anthropic

open ai how do they get out it's this rush to get out to the public markets if you want to slow things down you can you don't need to raise another 65 billion dollars at a 950 dollar valuation and then start kind of leaking that your s well or i guess actually actually filing your S-1. And, like, if this is truly a risk to humanity that can pose significant societal risks, and, like, wouldn't you stop? Why wouldn't you? Like, what do you think it is?

Do you think it's just, do you think Dario believes this? Do you think there's a large group of people that do believe this? But then the marketers and business folks and growth folks override them? Do you think this, I know, I mean, we go over this a lot, but where's your head right now as we get closer to the Anthropic IPO of what really is behind this messaging and what do people really believe? All right. First of all, I think that we should sort of tackle the core message here. And would you, before we go into what's going on within the company, would you agree at least that the AI coding tools from Anthropic have gotten way better and enabled people to do much more? six months ago, yes. I think the 4.8 was actually like kind of like a pretty big dud. Okay, but the line is pointing upward.

No, no, of course. But none of that. If whatever exists in the last one month existed in their line of sight six to eight months ago. If that was truly the case, one would hope that decent people would be. like, let's figure this out. Let's actually slow things down. Again, one to two months before your actual IPO to continue this, maybe we should have like, again, in terms of the actual qualification of risk of overall AI, I think it's just a separate conversation from this. For the sake of argument, you could argue. You could argue. Double argue.

Anyway, but one day I'll learn how to speak. Any statement that starts with that I'm excited for. That they did slow down to a degree by deciding not to release mythos to everybody. And a lot of this coding improvement has happened because of mythos internally. You're saying that, okay, so you're still, for the sake of an argument, you are arguing that that was the official slowdown. I don't know. Like, is it still raising this much money, pushing ahead with this intensity, like, organizationally,

like one decision? Maybe Mithos is this be all end all God model that like will destroy humanity and they have done. No one's saying that. No one's saying that. We'll break the fabric of society. Maybe it's not like, you know, so special. Maybe some of the latest opening high models.

can do similar things. But what it's trying to do is set this tone that's like, you know, if you can use this model to hack through things, we should be careful about the way we roll it out. No, I appreciate that idea. But still, then you don't roll it out, as we had to discuss. Like, then you don't have a big press announcement with named partners. And then now there's reporting that they're apparently working with the NSA. And like, again, they have the greatest PR and comps team in the world. in terms of like understanding which stories can actually make it into the press. Like, you wouldn't do that.

You would actually, in the background, coordinate and kind of work with everyone who is needed to to try to actually solve this problem. It wouldn't be a PR release with a large consortium named. That's why even that, if it were real, I still think it would be handled in a more in a different way. Like if it was real. I would not. How would you handle it?

You have a new... Only because... Go ahead. Go ahead. No, no. Let's say you got a new model. You suddenly, Alex is sitting there in a Brooklyn apartment. Has a bunch of a...

Your own colossus set up. And suddenly you train Cantorwitz 5.7. Suddenly you're like, oh, my God, this could... Has the potential to post-significant. societal risks, what do you do next? First of all, if I had a colossus set up in my Brooklyn apartment, I guarantee you the morning after I set it up, my wife would have every single one of those servers on the curb because we would finally have space in our apartment.

So I promised you the next day there would be racks and racks of your Rubens on the curb in Brooklyn. I promise you, it doesn't matter how much they're worth. Okay. Okay. So now, what would I do? And this is what I was going to say. Only because you're so myopic about this, wouldn't you do exactly what they're doing? So this is the argument, right? The argument is if you are, why keep pushing forward if you feel the technology is potentially dangerous? You keep pushing forward because you feel the technology, while you feel the technology is dangerous, because if you are building the leading models, you actually have a say, in the way that the deployment works. People will listen to you. If Anthropic was like meta right now

and had like, you know, whatever their latest, you know, model is that they're delaying, by the way. And they said, hey, you know, guys in the front, maybe slow it down. No one would listen to them. The argument is you only really have influence on the way this works if you are leading.

And the way that you do that is you're loud about it. So you are taking the I alone can fix it approach. I'm not, first of all, I'm not taking this approach. This is you. This is you and your Colossus and you have just made a significant breakthrough. Don't put this on me. I'm just trying to talk you through the other side of this heart.

I like how in this thought experiment you have just turned into this world, like this James Bondian villain. But continue. No, I promise you. My real world experience of this would be waking up in a Tony Soprano bathrobe and walking out getting the news. paper and seeing like rocks and racks of grace black walls next to the fire hydrant. I promise you. Okay. And that was actually that slows down the destruction of humanity and that's why she did it for

you. That's right. So thank you for that. Thank you. But seriously. Yeah. I mean, what is your, I just want to get your response to that argument that you, you stay in the lead to people listen to you and you're loud because you want to have influence over the

process. My response would be they have been saying this since the beginning. And I do think there was a moment that they might have believed it, but no matter who is in the lead, Open AI and Anthropic, Sam and Dario, but now the entire organizations, actually Open AI has pulled back from it. You can see pretty purposefully. But this idea that this technology is so incredibly dangerous, yet I will continue to push ahead at full steam has always been kind of the narrative and it's

helped them tremendously. I mean, have you ever heard Sundar say that? He said it's fire, but, uh, which would be funny hearing. No, you haven't. But obviously Sundar's perspective is going to be different than theirs. Why? Why? Well, I guess maybe obviously is the wrong word, but it, Google has always been more capitalistic. What don't you say? I mean, they removed don't be evil, right? Like they started that way. They became a normal company.

Oh, you're buying into it, Alex. You're buying it. Raising $65 billion at a $950 billion valuation is about as capitalistic as it gets. No, okay. No, you're right. You're right. And let me now, I was saving this, but you've now sort of forced me to spit it out. Let's do it.

Speaking of Google, the most. Okay, I learned a lot from Jeff Hinton when we sat down and that interview was on the feed from the Wednesday show. The thing that will probably stay with me the longest from Hinton is not him saying AI is conscious, although I thought that was really interesting. It was our discussion about safety and particularly whether Anthropic would be able to pursue its safety mission while on the public market. And as Hinton points out, when you IPO, you have a legal obligation to your shareholders to, to maximize profit and not to maximize AI safety. And I just don't see how it adds up where Anthropic continues to pursue the safety mission

while it's legally required to maximize profit for shareholders. Yeah, but you also have a legal obligation to not destroy humanity. I mean, again, like, that's a hyperbolic statement, but like, I mean- Who side are you on, Roy? No, no, no, no, no, but it's like, again, if you can be, Philip Morris turned all

TRIA is a public company, but still has to abide by, you know, I guess those are actual regulations and they're not self-imposed, but like, still, we are, in 2026, the whole corporations is good. Everyone kind of move past that, but like, still, I think if you have a fundamental technology that can pose significant societal risks, investors aren't, shouldn't like that, if that were true. Okay, let's just give a risk. That presents a risk. Let's give an example. Let's talk through an example. All right, Anthropic is public. It has mythos. Open AI has just released like a dynamic coding model, right?

And, you know, yes, there's a cost-benefit analysis. Yes, it might lead to more hacks, but it also will end up, you know, providing Anthropic with the ability to grow faster than Open AI and head off this threat. couldn't there be shareholder lawsuits if it decides to hold back mythos because there's, you know, it's not an end of the world situation. It's, okay, there's a few more cyber hacks that we want to be mindful about. It becomes much more difficult to do that when you're on the public market. Are you saying that Open AI presented the risks of mythos as just a few more cyberhacks? No, you're saying anthropic?

Sorry, anthropic, anthropic, yeah. Is just a few more cyberhacks, not a big deal. maybe your square space site goes down a couple of times. It's up to, first of all, it's up to the shareholders. And second, it becomes, wouldn't you agree, it becomes much harder to hold something like that back if you're on the public market? That's all I'm saying.

I think it brings a lot more attention to it. And actually, the good thing is it would actually force more transparency around the reality of the situation. You just took my side. Is that's why I was trying to say the whole show? No, but I don't think it forces them. It adds, I don't think the legality of a shareholder lawsuit, because Anthropic in a competitive battle does not release the model that actually they have marketed themselves presents significant risks

and had to, like, national security, like at every level, I think shareholders, the idea that they would sue that you did not release that to the general public when a competitor was being careless and callous, I don't think would happen. I don't know. And nowadays, anything's possible in the financial markets. So let me see if I can, I think there is consistency to your position here. And I think I reacted to something different in terms of your transparency is the transparency from the company, not transparency in the industry. Fair enough. Is your position, that right now, Anthropic telling us about all these, you know, models. And by the way, I mean, we've sort of kind of talked over it,

but they did then, after saying that the models could recursively self-improve, say it might be a good idea for frontier building to pause for a bit. Okay. So is your position that them doing this, building and saying this, is actually marketing. And actually, they are a capitalist company, you know, as everybody else. Yeah.

And don't believe that stuff. It's marketing. And when they go public, being public will still, you know, even though they are, you know, this, a company that does good marketing and is seeking and is trying to maximize profit share, maximize shareholder value and doesn't have a special thing about privacy, a special thing about safety, they'll still be required to act with safety in mind. So it's not inconsistent. Yes.

I think so. I think we've aligned again. Well, I'm just repeating your point. I will say it's a valid argument. Are you agree? Okay, okay. I'm not ready to agree just yet, but I think it's a valid argument. But I want to bring it back to the beginning.

Do you think the average anthropic employee is genuinely believes they own and like operate technology? And in their own words, that pose unprecedented cybersecurity risks severe enough to cause widespread disruption to the global financial system and critical infrastructure. A little more dramatic than a few cyber hacks. Okay. I have this. Let's end with this. I have this actually from the Wall Street Journal story, a quote from one of our favorite sources, a friend of the program, I guess he's been on once, Ethan Mollock, a professor at the University of Pennsylvania's Wharton School and an influential scholar on AI transformation. Here's what he told the Wall Street Journal. While some anthropic

critics see fluff and marketing in their safety pronouncements, many within the company are true believers. AI labs are a mix of things, said Malik. There is a trillion dollar company with all the normal trillion dollar company stuff, like marketing teams and lawyers. There is a core of researchers who are just building the next models. And then there is a set of people who are philosopher kings and concerned about the future and what comes next.

And they're all in conflict with each other at times. I think that's about as good an explanation as you're going to get. we just talked about this we've been talking about this for months Ethan just he just made it all make sense to me is this coexistence book going to be about this concept this idea oh so he has a book AI coexistence coming out in the fall according to the story I don't think it's going to be about how these people coexist with an anthropic I think it's going to be

about us coexisting with AI but I can bring him on the show yeah yeah he'll just co-existence Existence. Let's coexist together. We shall. Oh, speaking of coexistence. Well, that was a beautiful segue. Can Microsoft an open AI coexist? Microsoft has been on like the weirdest press tour this week. I don't know if you've paid attention to it. But they had this build event where they announced a bunch of their AI announcements and the core thing was like, hey, look, you know, we are, we're not wedded to open AI anymore so we can just go do our stuff. And we're going to build like the best AI models. But a lot of

it was like them talking about like how like we know i mean i guess you want to acknowledge it you're behind just like google did recently with the coding apps but they really got taken to the cleaners in a couple of these pieces my favorite line was one in the verge let's see for years Microsoft AI business leaned hard on its early and exclusive partnership with open AI but the drama-filled marriage slowly devolved into a situation ship and the pair effectively separated in late april though Microsoft is still Open AI's primary cloud partner for now. This year's build had the vibe of a freshly single divorcee posting a thirst trap on Instagram. It's always fun to be at developer conferences and great times have changed. Microsoft CEO Satya Nadella said on stage Tuesday adding that events like

this are about coming to grips with the new opportunity. And to do that, Microsoft has basically said, all right, we are ready to go and compete head to head with Open AI. The goal is to prove that we can become one of the top four labs in the world, Mustafa Suleiman, the head of one of Microsoft AI divisions told The Verge. He said, there's three labs that matter. Google DeepMind, open AI, an anthropic. We are not one of them at the moment, and that's always been my intention. I just love that meta is not on there, even though, I mean, we kind of know it, but still, it's always interesting to see it place that definitively. I've been thinking about this, so The same way Apple solely by owning the end-user real estate with the iPhone has an opportunity.

Microsoft certainly, I mean, we hear about this all the time. Everyone has co-pilot, tens of millions, hundreds, millions, people, whatever it is. Like, if they figure it out it's interesting, do you think there's a world where Microsoft and Apple a year from now are interesting in this? I'm not even going to say are winning, but even the way Google within a year, the span of a year became a player, do you think Microsoft has a chance? Do you think Microsoft to end? Definitely. Yeah. Sure. Why not? I mean, they have, like you said, they have office. They have access to opening eyes IP. I don't think they're like, well, here's a better question. I mean, like, First of all, okay, let me answer your question,

then I'll ask my question to sort of end this. So I think there's a chance. And they have the, like you said, the products. I think there's a chance for Apple. They have the products. If you're Microsoft, why do you care about building foundational AI models? I seriously don't get it.

I think it's because of Suleiman. I think it's like this is one of those things organizationally. Like, he's there. He's not there to like build a new co-pilot product feature. And my favorite still is in the most Microsoft way, which kind of Google seems to have moved past. Like, they have this new thing, Microsoft IQ, but there's also work IQ, then there's fabric IQ, then there's foundry IQ, then there's Web IQ. Like, just the way they kind of roll out and name products is comical at times.

But I think they, like, that's the stuff that that layer is not interesting. Like, he was brought on and he has power within the organization to be a player. And I think it's that simple. Do you think it's... But I want to dig a level deeper on this. All right. So he was brought on to do this. Why strategically is there a reason for Microsoft to do this?

Can you think of... I don't think there is. In fact, I think the Apple... So I can tell you, model routing has become like so dominant in conversations in my world now. Like it went from, you know, like just there's... You have to have the latest frontier model and they're all encompassing. And within the span of two months, everyone is like find the right model for the right problem, model cost optimization, tokenomics.

So if I'm Microsoft, I actually agree. I don't think that's the right approach. I think, like, you know what? Let them all battle it out. Apple over there is going to suddenly make Syria powerhouse because it's going to be called the best model at the time. and we're all going to forget about Bella Ramsey's ad, and maybe Microsoft has a chance too. Okay.

Let me give one reason why they might do it. Saty Nadal is speaking to Ben Thompson in Stratere. So Ben Thompson tells Satyndela that OpenAI is a major tenant on Azure. He goes, they're a major tenant, but let's face it, face it, Anthropic overtime or Open AI overtime will build their own. it makes sense. Pretty sure that's talking about a cloud service. And OpenAI has talked about building its own cloud service.

And so ultimately, if you're Microsoft, you're going to compete with OpenAI. So Open AI can build the servers, which you have for Azure, and then put the AI model on top of them. The only way to compete with that is to have your own foundational models. Maybe that's the reason. Okay. I mean, I still don't buy it, though. But. Why not?

I don't know. Like, if Open AI Cloud, Open. Open AI makes Open AI AI cloud and you can't answer that. Right. Then maybe you do want to build your own models. That's a big if. That is a big. That is a big if.

I mean, it's a lot of ambition. It is actually, but Codina-Dadella believes it's a possibility. Wait, that is actually interesting because it's funny like heading into IPO season, like hot IPO summer with anthropic Open AI and everybody, SpaceX. No one has actually talked about the AI cloud in terms of Open A. potential business model. That was like a passing comment from Sam. Like, do you think Satya really believes that or has some information?

I don't know. Like the fact that he's even addressing that is kind of shocking to me. Very interesting. Yeah. Sorry, I had that. I put that in the duck and forgot about it. But that to me is what you might be potential. That's a story.

That's a story. Funny enough, none of the report. I didn't see any reporters outside and, you know, bring that up and make that connection. I mean, I guess Thompson was the best at getting it out. but still like the conversation. What does Satya know? What is Satya? I mean, he's got to know something.

He's got to know something. He has all their IP until 2032. Yeah. All right. Well, folks, forgive my voice this week. I just completely blew it out at the Beer Authority next to Port Authority on 42nd Street in New York. On Wednesday night, watching the next beat the Spurs. And we're about to head into Friday night.

And I think I'm going to let what's left of it. go. So maybe it's a good thing I'm not going to be on site at WWDC. I won't be able to speak. I, uh, I'm sitting here looking down at Madison Square Garden from the 18th floor of a Penn Plaza. And it's, uh, it's, there's a lot of people out there, even though they're not playing here. There's blue and, uh, orange as far as the eye can see. Are you, you're a Celtics fan? I am. I'm rooting for the Knicks right now. I'm out of New York sports teams. Like, there's no part of me that could ever come around to the Yankees, giants, or jets. But I don't know, the Knicks, especially this team, I mean, you can't hate. And I'm, they've, they've, I'm not saying I'm

bandwagoning fully, but I'm certainly rooting for them. But as a general fan of the NBA, I mean, these are the, I can't root against either of these teams. And it's just good to watch. So I'm enjoying. Such a fun story. Yeah. Yeah. So that's, that's, I don't know about you. That's my weekend tonight yell at the TV for three hours straight tomorrow wake up get the newspaper and see a whole stack of data center servers on the curb Sunday flight at San Francisco only to not be led into WWDC I'm excited I think uh wait did you hold on I'm just gonna there was that story that Invidia is gonna pay people to like build data centers in their house or something I saw that

sign us up Jensen we're ready no we're ready actually no Alex is not ready I would you would find my cold body on the sidewalk in the morning. I put one of those in the backyard that we don't have. Yeah, I think New York apartment holders are probably last on that list with the media. Yeah, we can pass on the backyard server. All right, Ron John. Have a great weekend. See you next week.

And folks, Ronan and I are going to be doing the show live at the Big Technology Summit on June 18th. I think it's probably the last time I'll mention it until you hear some of the episodes because we're going to post them on the feed. It's going to be in San Francisco at the commonwealth club summit at bigtechnology.com. We're just about sold out. Thank you to everybody who signed up. It's been pretty overwhelming the response that we've gotten. And we're going to have one hell of the day out there in San Francisco.

So I'm stoked to be doing that with you, Ron John. I'm excited. And yeah, it's going to be great. And I think we'll leave it there. So thanks everybody for listening and watching. And we'll see you next time on Big Technology Podcast.