Right.
There's misinformation clearing order inbound.
Let's just roll. We are surrounded by journalists. Hold your position. Come, get up. Trust the experts here.
We are actually founder mode. Five good.
I see multiple journalists on the horizon. Stand by.
UAV online. Blaze. Double blaze. Triple blades. Double kill. Fight. Wrong. Cook is team deathmatch. We are experts. Triple blades. That's just wrong. Right.
Marky clearing order inbound.
Get out. We are surrounded by journalists. Hold to position. Strike one.
Strike two.
Activate golden retriever mode. Marky clearing order inbound. Five.
I see multiple journalists on the horizon in Denmark. Founder.
You're watching TVPN. Today is Wednesday, June 10th, 2026. We are live from the TVP Ultradome, the temple of technology, the fortress of finance, the capital of capital. Let me tell you about ramp.com. Baby, time is money. Save both. Easy. Use corporate cards, bill pay, accounting, and a whole lot more. Ramp, Ramp. You go to Ramp.
You want some accounting help? They're going to say yes and thank you for using us.
And what else?
Two major stories. A bunch of major stories, but the big one: Facebook got a movie, another movie, actually a third movie if you count the documentary. There's been The Social Network, The Social Dilemma, and now The Social Reckoning. Only two of those are movies. The Social Dilemma is a documentary, but it's very interesting. And Anthropic got a fable, and there's been a ton of reaction. Ben Thompson was writing about it. The Wall Street Journal, like the announcement should have been like the really good model launches and people are amazed with what it can do. But they put the title 'Anthropic Puts Curbs on AI Models.' So, well, I'm sure you've seen this on the timeline, but we'll take you through some of the debates over how restricted the model is in certain areas. Bio is a big one. Cyber is another one. There's a bunch of funny examples. There's a bunch of reasonable arguments for doing this type of stuff. So we'll go through all of it.
Can we talk about who's joining the show today?
Absolutely. We got a bunch of great guests.
What a lineup. Hey Clicky founder Farza Majjid is joining. You've probably seen his viral demo last week. We're very excited to talk to him about computer use, potential consumer AI applications, what he's doing to build hands-free AI voice assistants. The CEO of Snowflake is coming on. The CEO of Cloudflare is coming on. Vinod Khosla from Khosla Ventures is coming on. And then Bret Taylor. Bret Taylor didn't even make the cut-off on today's lineup. What's going on here?
Brutal. And of course, we have Marki Wagner.
Coming out announcing emerging from stealth with her new company.
So I want to kick it off with the Social Network sequel. I want to watch the trailer for The Social Reckoning because Jeremy Strong is playing Mark Zuckerberg. I've heard people say, 'Hey, he looks like he looks like Zuck.' Like the styling is there.
Henry says, 'No, Doug from Semi Analysis.' Oh, we got to get him on.
And see if he's available.
Not that we have any time, but we'll make
Well, we'll make time for Doug anytime. So the one critique is Jeremy Strong looks a little bit older than Mark Zuckerberg did in 2011 because he is
Calling him or it just doesn't... it will color the movie a little bit in the sense that the film is telling the story of this very difficult whistleblower situation. It's different when it looks like a mature adult versus someone who was young and running into these problems for the first time.
Yeah, he's joining. We got the Cloudflare
12:30 Pacific. Get ready. We're going to go through Cloudflare. They have exciting news. I think it's public already, but I don't want to blow it up if it's going to be announced later today. Anyway, I want to react to the trailer. So let's pull up the trailer.
Listen, before I go on, I want to make something clear.
I have a hunch you're not a fan of Facebook, but I am. I am here to help Facebook, not hurt it. Okay. You send me a message. What would you like to talk about? The chairman gave a session to order. You'll read your opening statement, which we'll skip past for now. That's a separate session. And we'll move to witness questioning. Spell your name and state your current occupation for the record.
So, is it not this time?
Cerperg.
Let's see. And your occupation?
Zuckerberg. I'm a professional defender. Co-produced and directed by groups where 30% of content hits multiple risk factors. Hang on. I don't piece to the social. Are you a tech reporter?
Is this my guys are counting on the next round of professional testimony to make you mark? I'm happy to lend a hand, but I think you do. You wrote broken code. He was a journalist at the Wall Street Journal. He is a journalist at the Wall Street Journal and has broken a number of global news stories in our lives including the Facebook files. The fire hose of bad information you are injecting into the air supply jet power. I'm a free speech absolutist. I'm not the one who's lying and I'm not stopping them from seeing someone who is.
It's not a bad Zuck impression. He's got the voice tone pretty down.
Got worse as a result of time spent on the platform. Senior leadership knows and is doing nothing. I know there are easier enemies to make. The mafia would be an easier enemy to make. So what would you need to stand up the story? The internal documents. This is a material violation of my NDA. We're twice as big as the biggest country on earth. We're not frightened of Congress. We're post-government around here. Please. Twice as big as the biggest country on earth. We have 102 hours to get from a user standpoint.
I don't want to be made an example of by a guy with unlimited resources. H, I promise you, is imminent.
Enough. People around here understand that when I say no, that's the end of the debate. I'm not 2 years out of a dorm room anymore. Charlie, look around.
Man, if they were trying to go as inspirational as The Social Network, they really missed the mark there. What happened?
Yeah. So their intention was to try to fire up the next generation of entrepreneurs. That seems like they missed the mark. Inspire them, uplift them.
You can build any business out of your dorm room.
Yep. And also it feels like if the goal of this film is to really take the viewer inside Meta and superintelligence and what's going on with Nat Friedman, Daniel Gross, Alex Wang, it feels like they dropped the ball potentially. It's sort of crazy.
Yeah. Like I didn't see anyone that even resembled Yan LeCun.
That is crazy to not highlight that and what's going on over there. The launch of Llama, the Llama 4 moment, could have been the crescendo. Behemoth and all the drama around that. You want drama. You want Hollywood drama. You want Oscar bait. Llama 4 Behemoth. That story is going to put butts in seats.
That's right. More seriously, it's hard for me. It's going to be really hard to not look at Jeremy Strong and OC Kendall. Yeah, because you're a Succession guy. So this is the sequel to The Social Network and instead of chronicling the birth of Facebook, it's the story based on the 2021 Facebook leak by whistleblower Frances Haugen. It's going to be dramatic. It's not going to make people like Facebook more, and it's probably going to make Americans even more distrustful of tech. What's interesting here is that if you stop a random tech person on the street and ask them, 'What is The Social Reckoning about?' They might say Cambridge Analytica because that was another drama moment. I was sort of like, is it Cambridge Analytica? Is it the Haugen thing? A lot of people don't even remember what the whistleblower story was. Little refresher there. It was an internal document leak. The Facebook files showed that Facebook internal employees were aware of harmful societal effects from its platforms, yet persisted in prioritizing profit over addressing these harms. Now, Nikita Bier chimed in, who now works at X, a rival platform to Threads, and said, 'Zuck makes a lot of mistakes, but this isn't one of them. Meta literally had multiple teams of million-dollar-per-year engineers working on teen mental health, and they had the agency to override big product decisions. They're probably a thorn in Nikita's side because he's trying to maximize click-through rates, maximize attention, and he's getting pushback from guardrails here.' He says the story this movie is about is actually a product manager who didn't get a promotion. So that's the pushback.
The only thing with that argument is you could make the same argument like the cigarette company has million-dollar doctors and researchers focused on making sure cigarettes are as healthy as possible, right?
Yeah, especially with the new fertility data, I think there's a new cycle brewing of exactly how bad is social media. I still don't really buy the addiction thing just because I'm familiar with addiction defined in nicotine, which is extremely addictive. It's chemical and you have cravings. If you forget your phone, do you have the same cravings? It's sort of different. But there is a very serious discussion going on around social media and its effects and this fits right in.
The phone addiction thing is very real. I was at an event last night and they required everybody to put their phones in these locked bags. You could go outside at any point.
I didn't feel sick, but I noticed probably 20 times throughout the night I was thinking of going for my phone.
I didn't start having physical withdrawal symptoms.
But the craving was there. I just kept thinking about it. It was like phone noise.
Oh, this is real. This is real.
I was having some phone noise. Okay. So that's legitimate. The question that I had really quickly is what is the impact on Meta? It's a one-and-a-half trillion dollar company. They're trying to raise more equity for the AI buildout, hire people. Is this good, bad? What does it look like over the short, medium, long term? Short term, I think this was amazing timing. They really got lucky because the fact that this trailer dropped the same time as Anthropic's Fable 5 was incredibly fortuitous because Fable just took over the timeline. No one was really talking about this in tech. Tech insiders won't really see or talk about The Social Reckoning this week. Case in point, I put it second in the newsletter when I wrote it up. Medium-term, I think Jeremy Strong is going to drag Mark into the bad group of AI leaders while he's on his Oscar tour. There was a version of Facebook strategy, a Meta strategy that's just like, hey, we're like Amazon, we're like Microsoft. We have bets in AI, but we're not deciding the frontier. Obviously Mark has made the decision to hire some of the greatest researchers, invest very aggressively and this puts him at the center of the conversation around what will the effect of AI be on our society, on kids? Should kids be talking to LLMs? Psychosis? All these different things will come up. It primes the pumps for the next wave of people went into the Meta app and had a bad time and now we got to talk about that.
What do you think Mark is going to be more annoyed about this week, the fact that SpaceX is going to be valued at meaningfully higher than Meta or this trailer?
It's actually a tough one. I don't know. I think the real one is Fable 5 because if Meta was using a ton of Anthropic models and then the new model comes out and says you specifically can't use it in AGI, the most important initiative in the company, that's a rough thing. You've been working with this company to develop my models for my family of apps and now I can't use them. Meta is spending billions of dollars a year with Anthropic. They have been for a while.
They will spend billions of dollars this year. The question is if Anthropic does allow Meta to use their models for Meta's AI research, what does it say about what Anthropic thinks about Meta's potential in AI?
Secretly love ads this whole time. I don't know.
To me, it would be them not feeling like they had an ability to get to the frontier.
Totally. And that's an ace up their sleeve at any time. They can always just take the guardrails off and be like, 'Hey, more AI, more intelligence on demand for everyone.' I think the fallout of the medium-term, Jeremy Strong's going to go on this press tour for the Oscar and he's going to have a bunch of emotional sound bites and also be viral because it's funny to see him doing this impression. He's already gone viral for his Mark Zuckerberg impression in the past. I wouldn't be surprised to see the next version of a Bernie Sanders press release about AI highlight Dario, Sam, and Zuck instead of the historical order which is Sam, Dario, and Demis. It's always based on perceived industry power and negativity around personal brand. They have to have a scary quote and they also have to have a lot of power in the industry. Zuck's power in the industry is rising and with this movie, there's more ways to take shots at him. You can say, 'Look at what he did with the Facebook files.' People don't remember the details, but after they see this movie, you can say this guy's also building crazy data centers and if he handles the AI thing like that, it's going to be bad. So that's a risk in the medium term. Long term, I don't think The Social Reckoning is going to matter all that much for Meta. People will complain about Meta data centers, AI plans.
Why did they not get Jeremy Strong a wig?
They did. He is playing an older. Remember when he had the Caesar haircut? So it's a different time. People will complain about all this stuff, but they'll do it on Meta's family of apps. I don't think there will be churn. Advertisers won't pull out. It's impossible to pull out of Meta because it drives up the return on ad spend for smaller companies. A bunch of small businesses will jump in and say, 'Great, some big company is boycotting Meta, I'm Ridge Wallet, I'm going to jump in.'
Advertisers have tried to boycott it.
You can't on Facebook. It's impossible. The business is just too strong. I also don't think Mark will be singled out by a regulatory hammer. If there's a data center ban, I don't think it's going to be uniquely focused on Meta. Only Dario or Dario and Sam have true scapegoat risk because they run pure play labs and have been so noisy about AI and they have the biggest revenues in the category. There might be data center regulation, but it won't unfairly target Meta. So anyway, that's my Social Reckoning take. Let's move on to Fable, which launched yesterday. The model seems incredibly impressive. I've been seeing these vibe-coded games that look really good. They hit well on social media because you take a video of them and show the example. People just sort of believe it. It could be embellished, but in general these feel pretty solid. Of course, making a great game requires great mechanics and multiple levels. A single demo of a forest is not quite there, but really useful. We'll see a bunch more examples of games as memes, simulators. These things are going to get easier to build. That's really exciting. There was a bunch of debate on the timeline yesterday about the latest model, Fable 5. The first Mythos class model is remarkably good at long horizon tasks like software development and knowledge work, but rejects requests related to biology, cyber security, and frontier LLM development. Interestingly, I haven't seen anyone share rejections around anything else. Did they remember to reject 'build me a nuke'? Because I haven't seen anyone try that. It would be very funny if they did not get around to that. I think a lot of those other things that you should reject have been ironed out in previous iterations. Saying a slur was a big one for a while. Or saying something rude, pausing a chat, shutting down a conversation versus switching you to a less performant model.
Yeah. And it creates this screenshot that went viral pretty much continuously yesterday. This aligns with Anthropic's focus on safety, but as many people have pointed out, it's also just good business. You don't want competitors using your products to create competitors. You also don't want financial liability or negative headlines from bad actors using your models for nefarious purposes. Ben Thompson called it true alignment. The take-safety-seriously culture aligns with business value creation, which is very rare. Oftentimes, the 'be good or your culture' limits what you can do and hurts your business, but it's something you do in favor of brand. Apple would probably be more profitable if they were using diesel generators for all their data centers. They went clean energy because they wanted to have an environmental brand and over the long term it's helped them, but in the short term it's been rough.
Inexpensive. So Ben Thompson writes, 'What is so fascinating about Anthropic, however, is that while I'm sure some executives of the company are thinking this way, I also totally believe that the employee base broadly also happen to believe that they are doing the right thing.' It's fascinating to observe. Me, the rational business analyst, sees a hard-nosed but understandable decision to cut off would-be competitors. Anthropic employees and advocates, the true believers, see a regrettable but understandable safety decision that ensures that responsible and thoughtful people themselves will be the ones guiding our AGI future. This is true alignment and it's an incredible accomplishment. Facebook has tussled with this a bunch. We already talked about that. To be clear, Fable 5's rejection threshold really does feel way too low from what people are saying. There are tons of examples on the timeline of a biologist just saying hi to the model and getting kicked down to Opus. I saw someone just said 'cyber' with the devil horns emoji, and it's like, 'We can't, we're not going further.' Bringing down a broad hammer, you can always dial it back over time. But every rejection is an implicit invitation to hop on the phone with an Anthropic sales rep and get on the Mythos enterprise plan. That's where the real dollars are. The timeline is unhappy because the idea of democratizing science and technology is alluring, but the pool of dollars available from all the biohackers in the world probably isn't close to the budgets available from big pharma. So you're in this rational business analyst situation and you fail to see how this is damaging except to the hacker community. The real tricky part is how frontier AI research is handled. Instead of outright rejecting the query and bumping the user down to Opus, the model appears to answer but quietly gives a degraded answer. This was disclosed in the model card, which is interesting.
Business not disclosed to the user in the product while they're paying for it, which is a different path than bio and cyber. If you go LLM frontier research devil horns, it will actually give you an answer apparently. It won't bump you down immediately. But it does not disclose that, which is odd. Outright rejecting requests for AI research and saying, 'Hey user, sorry this model doesn't work for that type of project. Please use another model or contact sales if you want help with this,' would have been much more in line with the bio and cyber security strategies. They also could have not disclosed it. It's possible that they just didn't need to disclose this at all. They could have just released a model that was intentionally nerfed on AI research. It would have shown up in the benchmarks because people would have benchmarked it on some sort of AI research bench and been like, 'Oh weird, it's really good at all these other things, but it's bad at LLM research.' That might have been a bit of a brand hit, but users might never know that the model was intentionally degraded around this category of work. That leaves this third more worrying position: intentional degradation without disclosure in the model card. There is no evidence of this, but it's possible that other workflows might be nerfed and there's no law or convention around disclosure. Again, maybe good business, but a weird situation to be in. Probably bullish for evals if you're building a business on top of a big lab. A legal AI company will want to be really sure that the models they're using aren't degrading unexpectedly and not telling them. It's different if you're a bio researcher using this and you know the model was never intended for you, but what you don't want is to be using it and it's now giving you misleading results without telling you. That would be an odd outcome that they'll probably address in the near future.
Yeah, that aspect triggered Dean Ball. He said, 'My last observation re: Anthropic secret sabotage safety policy is that it undermines actually good safety policy.' How? First, it is very plausible to describe this as anti-competitive behavior. Even if you are maximally sympathetic to Anthropic here, you must admit this. And this behavior is being justified in the name of AI safety. If you believe, as I and many Anthropic staff do, that it may end up being critically important to relax antitrust enforcement so that frontier labs can cooperate and collaborate on some areas of AI safety, Anthropic just undermined the case for that in a large way. Overall, this massively and profoundly raises the status of the argument that AI safety has been hyped to justify monopolistic behavior by labs. I continue to believe that AI safety is a real and serious issue that is growing in importance rather than diminishing. If you agree with me, this incident is a setback, maybe a serious one. Third, he says, as I have observed elsewhere, Anthropic's official corporate policy is structurally identical to the fact pattern alleged against them by the Department of War. I still think DoW acted both falsely and wrongly in that fight, but it is no longer possible to defend Anthropic with a full throat after this incident. This raises the case for heavier-handed regulations. Anthropic is making an awfully good case here that their product ought to be treated as utilities and thus their alignment practices should be a matter of public policy rather than private property. I am starkly opposed to this sort of state power grab, but Anthropic is doing more to justify it than anyone else. Thus, significant damage has been done to a community and entire approach to AI governance. It was done unilaterally by Anthropic, likely motivated largely by self-interest and justified within the internal psychology of the firm through the lens of safety. I suspect this is fixable in the economic and legal senses, but I fear that trust has just been broken and the goodwill extinguished will take very much time to repair. And just to level, he wrote the AI Action Plan, but also came out very publicly in support of Anthropic during the conflict with the Department of War, saying that the Department of War is completely overstepping by pushing towards supply chain risk designation, putting pressure on Anthropic for not wanting to work with the government in that particular way. Obviously there's a whole bunch of new data that's been released and that conversation has evolved. He doesn't strike me as some crazy hater. Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web apps, servers, databases, and more while Railway automatically takes care of scaling, monitoring, and security. Doug O'Laflin was also sort of mixed results on Fable. He says when it works it's brilliant, but the unilateral guardrails make me frustrated beyond all belief. He has a folder of health information with 100 days of aura health data. This is a very logical thing that people would do with Codex or Claude Code. Gather up all your health data. I actually talked to a very prominent person in tech about how they got into vibe coding and CLIs. One of the first things they did was reanalyze some Whoop data, reanalyzed some health data and correctly detected sleep apnea and went and got it treated. That is not the cure for cancer, but that's incredibly awesome. Who knows, maybe the thing that you detect early could lead to an increased risk of cancer in the future and that would be a really big win for our society. Little bit tricky there. He says there's a private investment in the life sciences tool space. Guess what? It's not safe. Doing some code scanning for vulnerabilities? Not safe. I get the safety, but it feels incredibly out of touch for a group of a few thousand people making all total comp in the millions, telling me what is and isn't safe. Dario worried about inequality. I think he has to realize that he is the inequality. The unilateral gatekeeping feels wack as hell. I don't like it.
People are people are very
Doug O'Laflin is extremely optimistic about Anthropic. He recognizes the power of the models in the business and has been a very strong supporter, I feel like a strong user. He was one of the first people to admit to Claude Code psychosis. He's unhappy with the current situation, but these things are very tunable. I'm optimistic with more disclosures and more fine-tuning, we can get to the good outcome here.
Last one: 'Tell me about mitochondria, it's the powerhouse of the cell, right?' Chat pause, the chat pause is rough. It's a screenshot.
But maybe the harness is just saying, 'Sorry, if you don't know that the mitochondria is the powerhouse of the cell, it's not worth my time.'
That's the thing, you keep going back and forth.
Yeah, there are expensive questions. All of these screenshots are clearly from $100, $200 a month plans. The GPUs are on fire as they always are with any new model launch from a frontier company. It can make rational business sense. When everything aligns and you're like, 'Yeah, we're being a little bit too safe, but it's actually good for our business,' it's very easy to say yes to that stuff. The big question is where you get tested where something is a hard decision and it doesn't align with your business interest. That's always tough.
What's going on? Was there the new data retention policies as well? That also has to do with trying to
This one was funny because I assume that every tech company stores everything forever basically. I didn't realize that they don't, not for these enterprise use cases.
I know that the enterprises said that they wouldn't train on your data, but when I go into any chat app, I expect to have my data from a year ago still there. One of my main critiques and frustrations with these apps is that I want to be able to, instead of hitting the search bar, just go into the chat box and say, 'Hey, remember a couple months ago we were talking about CrowdStrike and I had you pull up some data. Can you just go refresh that, get the original thing?' A lot of times these apps are like, 'I don't really have access to all your chats,' but the chats are clearly saved. I understand people are upset about this, but it didn't fully clock for me why. I understand the training thing, but
They're very explicit. They're not using the data to train.
They're not. Okay, well then that's good. The steel man I believe was that you need to hold it for 30 days because companies, probably international companies, competitors, will set up a ton of shell corporations and send out pseudo-random queries at random times over random accounts through VPNs. They will triangulate something that is useful to distill or learn from the model. By keeping it all, you can now run analyses and look at, wait, is there a pattern where 25 different accounts that seem completely unrelated all seem to be triangulating the same question? That seems reasonable to me. But stay out of my data. I'm built different. I'm not distilling anything. Anyway.
Let me tell you about Shopify. Shopify is the commerce platform that lets you grow with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents. Speaking of AI agents, we have a founder, a company. I don't know if we should call it an AI agent. What are we calling it? Farza. Is this creating a new category? Introduce yourself. Tell us what you're building.
What's going on, guys? Thank you so much. Farza here working on Hey Clicky. It started as an AI teacher. Now it's an AI where you can essentially talk to your computer and it does whatever you want it to do.
It's personal superintelligence and you got there before the big man. I love it.
Somehow I feel like the big man doesn't even know what they got sometimes, you know. It takes the little guy to figure it out.
I love it. I'm always rooting for the little guy. Reintroduce the product because you went viral last week, but it feels like you've been working on this for a while. What's the nature of the team, the structure, the funding? What's actually building this? What's adoption been like? Where is the product today?
I started eight weeks ago. I just wanted to build an AI that could teach me Da Vinci while I'm actually using Da Vinci.
This is Da Vinci Resolve, the color grading and video editing suite that's free from Blackmagic Design. Got it.
Exactly. Da Vinci Resolve is a complicated program. One-hour YouTube videos weren't cutting it. Let me just build an AI that can help teach it to me while I could talk to my screen and I could learn.
I've been there. Put it out and I thought it was a really bad idea. One of my friends just decided to post it.
So I posted it and it went really big about eight weeks ago. After that I didn't keep working on it. But then the cool thing was I saw all these people, when you put something new out there, you have emergent behavior. When you let a human being talk to their screen, what did they start doing? They start doing a lot of crazy things, from watching anime with Clicky to learning Blender with Clicky. Because it's so easy to just talk to your screen.
I just want to hear about the unhobbling path. I imagine there's demand for these chat interfaces that have existed for years where you open an app and do the voice mode and have it read back to you. But when I think about using a computer, I think rearranging windows, I want anime over here, Da Vinci Resolve over here, a YouTube video over here. I want to be able to puppeteer the mouse and keyboard with full context. So I need to be taking screenshots every minute, every second, 30 times a second, 144 hertz. How are you getting the data in? What's actually under the hood? Because you need to process the voice, produce an output. It's surprising to see a harness break out so quickly.
It's actually simple. We use GPT Realtime upfront to give you the first layer, a really quick answer. But if you want a deeper thought process over the image, if you're in Da Vinci Resolve or some complicated software, we now use Fable 5 by default. Fable 5 is absolutely mind-blowing in terms of how precise it is on screen understanding. We use that now by default when you want to understand something on your screen. We only actually screenshot when you press a button.
But it's really crazy because we can still detect the program you're on. That's already really helpful. If I know you're on Notion for example and you're there for 10 minutes, I can ask you, 'Hey, what are you doing? Can I help you?' It's that new modality that's been really exciting because that's something that's just not possible with any chat interface.
It's more like a co-worker walking up, just being present, kind of like sometimes you don't need help until you're asked.
I kind of describe it as having a 23-year-old intern, a new grad, always watching over my shoulder, seeing patterns in what I'm doing and tapping me and saying, 'Can I do that for you?' If I had that, of course that would be awesome, but it'd be cool if everybody had that.
A lot of these frontier models are expensive. How are you thinking about the business model? It seems like you've built something useful. People should just pay 10% on top of whatever the metered rate is. But in consumer, subscription flat pricing is very popular. That introduces financial risk to you. You have to understand your costs and how they change. What do you think the business model will be as you grow the business?
I mean, believe it or not, like
People are very open to paying large amounts of money, like normal consumers, large amounts of money for having better access to these models. Sure.
Different interfaces. So right now we just charge a straight up $20 a month. Okay. But even then there's a limit. You know, for $20 a month you get 150 agents on Clickie. Sure. But even that's very specific because once you pass that number, that's when we start losing money. So everything's done in such a way where it's pretty cost effective. Also, just so you guys know, calling Sonnet or Opus or Fable is kind of cheap overall. The expensive part is agentic work; that's really expensive. Just to give you an idea, on GPT-5.5, if you tell Clickie right now to go and click 'add to cart' on Amazon, that costs 25 cents just by API. Yeah. I know because I'm seeing the cost myself. So there's a lot of tricks to kind of reduce that. But overall, there are so many tricks to making agents cheaper, more efficient, changing the model based on their quest.
And I didn't catch it. Have you raised for this, or haven't you?
In the process. There we go. There we go. If you're out there.
Yeah. But if you're in the process, that means you didn't launch this and be like, 'I'm down to be losing $5,000 a day.' You've been focused on the economics probably earlier than you would have been if you had raised money out the gates.
I mean, yeah. I'm never the type of guy that wants to be losing thousands of dollars a day. So we can stop it early. But that's the thing about AI. If you've built a lot of social apps in your life, costs technically don't matter based on the number of users coming in. But in my case, for every 10 new people I add, I'm either losing or making money depending on what they're doing. So I've got to think about it. So yeah, AI is accidentally making everybody better at business because it's kind of required.
So I'm thinking about the hilarious tradeoff in having an agent that applies a coupon code but winds up spending more money applying the coupon code than you save, because it's like, 'Oh yeah, we put in this, it saved you 24 cents, but the API call was 25 cents.' But how important to you will creating an ensemble of models that sort of maximize the parade of frontier be as you build this business? Because I'm imagining that any day now we're going to get somewhat useful on-device models on Mac from the new Siri models. The on-device models will probably be very limited in what they can do but useful for some things, and that's free for you. Then you'll have open-source, legacy models, GPT-4 class stuff, and then you might only call out to a Fable 5 on demand. So do you need to build your own router, or is there a tool that you'll be using? How important will that be from the interactions that you're seeing?
That's a great question. You're essentially asking about the harness. We just built our own, because technically this is all so early there's nothing off the shelf. You have to think deeply about it yourself and then build it yourself. So for example, let's say you're on Clickie right now and you say, 'Hey, Clickie, I want you to research the latest in the Iran war and put it in a Google Doc for me when you're done.' Should it hit Opus? Should it hit Codex? Should it hit... well, it actually depends. So we just route them ourselves. We use about four models right now underneath the hood.
And do you have a model that's doing the routing?
GPT real-time actually does the routing.
Yeah, that's something we figured out that I think not even the OpenAI team really knew, and they hit us up about it: it's a great router. It's so good at tool calling, which means it's really good at routing requests. So for example, we have this tool called Fable 5. If a user asks something that is a heavier pixel task, we call Fable 5. If it's more agentic work, we call GPT-5.5. It's all being done in the background, but that's the magic of the product right now. Most people who use our product have never used an agent before; they don't even know what Codex is. For them, it's magical: 'Oh, I'm just talking to my computer and it's doing the thing.' That's awesome. That's what they want.
How do you think about the tension between you being a generational media talent and then startups? Because I actually feel like this is our first time meeting, but I've seen your videos over many years. You're just really, really, really good at it. So much so that as somebody who's in media, I'm like, 'I kind of want you to just do that podcast.' And I don't say that normally. Normally it's the exact opposite: 'Hey, you shouldn't do this podcast. You should just work on your company.' And they obviously feed into each other really well. You have an edge; you can get attention for free by investing a few hours making a video. It's a very powerful skill set. But how do you think about that tension?
You know, it's funny. I've been making videos for like 15 plus years, and it was an overnight success the second time. I always see this as a thing I get to do for fun on the side while I get to do the main thing, which is actually build stuff for people. At the end of the day, I am painfully familiar with how bad of a business making videos, making movies, and making music is. I know how bad it is firsthand. But I also know more than anybody the power of getting in front of a billion people every single month when there's a good business engine underneath it powering something else. So I fully intend to do that. If I have this media ability, I'm glad I got it.
Yeah. Sales. Really scalable sales.
You've got to be good at sales. WWDC had some movement on interoperability in iOS. It still feels like the Clickie iOS app is on the longer term stuck in the walled garden for a while. But what I'm interested in is: do you see a really solid market because there's no DaVinci Resolve iOS app that I'm aware of, and certainly people don't... there are professional tools that only exist on a desktop or laptop. Are you interested in hacking on crazy workarounds to build something that maybe makes people a little upset in Cupertino but gets you into that more casual mobile space, or do you think you just want to focus on desktop pro-level apps?
I want it all. I want when you talk to your computer, whether this is your computer or your computer is this computer, I want to be an interaction layer on top of it. That being said, I have no interest in being an OS-level controller like what Apple is doing right now. For us, most of our customers and users are essentially connecting 15 integrations to Clickie, from G Suite to Dropbox, and doing work with it. So I'm much more interested in that reality where I can talk to my phone in the morning and say, 'Look through all my emails and send me a brief when you're done.' I'm interested in that stuff, which is stuff I don't think Apple is going to do. They won't connect your 15 places to their Apple account and do work with it.
How do you think the health of open ecosystems in desktop software is broadly? Because there are two ways you could update an Excel file: you can puppeteer the mouse, move over, click, type the cell, save, or you can just edit the underlying CSV file in raw text and then refresh the front end. I imagine one is way cheaper for you, so you want to lean into getting the MCP servers, the APIs, the file writing, the CLI interaction. But do you think companies will lean into that or fight it because they want you to stay mouse and keyboard at all times?
That's a good question. I don't think so, because at the end of the day, what Clickie is doing in the background... just so you know how it works, we literally took the Rust binary that's in Codex and packaged it with our app. So when you call the clicking agent, you're calling a sub-spawn of Codex. This is on purpose: I want to use the best model and the best thing possible. I don't think that's going to end up happening. In fact, it's better. For example, if you ask Clickie, 'How do I add a formula to this Google Sheet?' there are two answers. One is, 'Let me show you.' The second is, 'I can show you, but do you want me to do that for you?' I think we're going to start moving towards computers just saying, 'Can I just do this for you? I see you doing it, I can just do it.'
Well, good luck with the fundraise. I'm sure you'll be back on the show soon. Let us know when it closes. Incredibly excited to watch you win and have fun in the process.
Yeah, it's great to meet you.
Congratulations. We'll talk to you soon.
Later. Let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI. Own the data platform that powers it. And up next, we have Trent from Side Talk. He's a creator and co-founder of Side Talk. He's in the waiting room. We'll bring him into the Ultradome. He'll turn Nick's fandom into one of the internet's most recognizable sports media brands. How you doing? Take us through. How is New York?
Yo, it's crazy out here. I'm not going to lie. I was about to do this interview from my apartment, and then I realized, 'Has anyone done a TV interview live from the streets of New York?' You've got to go to the sidewalk. The founder of Side Talk.
Amazing. For those who don't know you, take us through your media empire. Take us through your strategy. Break it down for us.
Okay. So I have a little company called Side Talk. You might see the logo. Pardon the potential loss of voice from this Nick crowd. But we kind of run around New York City and let people do and say whatever they want into our microphone. We created the 'bing bong' trend years ago. We created these NYX videos. Unfortunately, there's a little bit of chaos outside of Madison Square Garden due to us. But we pretty much let people say what they want in New York, and it's fun and entertaining.
What's the secret to a good street interview?
You've got to keep it sharable and engaging. It sounds obvious, but why would you watch an interview of someone talking about something boring or low energy? What would you want to send to your friends? That's the principle.
So you ask people about language model diffusion and token pricing and all private credit?
Always, always, always. That's what we love talking about. That's what the people want to see. We're going to start talking about that outside the Knicks games soon.
I understand that a lot of the man-on-the-street interview format these days can be very prepped. There's a PR person pitching a successful business person. It looks candid, but in fact it's staged. Do you ever participate in that, or is everything candid? Is that part of the brand?
We're 100% natural on the streets. You don't know what to expect.
Yes, sir. Organic. It's really cool.
Then my follow-up is: if you aren't prepping and understanding who you're going to interact with, how do you vet? Where is the bar? If you see somebody walking with a baseball hat on and they probably don't want to be bothered, is there a social contract where you shouldn't? Is there a doom scenario where everyone in New York is asked 'what do you do for a living' 15 times when they get their morning coffee?
Honestly, it's a bit of an epidemic. I'm actually kind of scared I'm going to get pulled up on right now and asked what I do for a living or what song I'm listening to.
Oh, listen. That's another one.
The people of New York distinguish Side Talk a bit from that. They know we're not there to ask them something cookie-cutter. Now people come to us when they see the microphone, which is great. We just go with the flow.
How do you think about monetization? Short-form monetization is notoriously hard, but you have merch. Break down the business side of the business.
We are using the power and knowledge of these models to create value for customers. Our coding agent KCO helps people get value from data, and our product Snowflake Co-Work allows every employee to access valuable data. We think about being in the path of driving consumption and creating value.
So my last question: you said it's a bad time to be making two- and three-year plans. How far are you looking into the future? Do you have a one-year plan, a three-month plan?
As a public company you have to have forecasts, but more qualitatively, our overall strategy is very clear. Anytime anyone wants to do something with data, I want them to use agent tools created by Snowflake, like Coco, to do that. We want to have the best data platform in the world so when someone needs a historical view or an OLTP database, they think of Snowflake. They can use any coding agent. We also think about how to sell Co-Work to CROs and demonstrate how AI can transform sales organizations. There'll be a lot of details, but that's the path we're headed in: AI leveraging the full power of data.
Well, congratulations. Thank you so much for taking the time.
Yeah, great to meet you and congrats to the team on all the momentum.
Yes. Stay strong in the token path.
We love to see it. We'll talk to you soon. Have a good day.
Goodbye. Let me tell you about agents meet the canvas. Your AI agents can now create and modify your Figma files with design system context. Apologies for the technical issues. A lot of people are asking if this is AI. There was some type of clip.
I've never seen that exact technical failure before where it cuts out entirely. We've heard scratchy audio and degraded video, but the video was crystal clear while the audio cut in and out. I think we got a lot of good stuff. Coco should have been called Frosty Snowflake.
Probably intellectual property.
Citadel Securities, where I interned, wrote a post on tokconomics. They argued that agentic and complex workflows delivered by frontier models would be expensive to run, constrained by physical bottlenecks, and vulnerable to unrealistic expectations. They said adoption is becoming less about what frontier models can do and more about the price and scarcity of inputs like compute, power, cooling, memory, bandwidth, and inference budgets. They see growing signs of a bifurcation in frontier versus everyday AI usage, with simpler models becoming more cost-effective until physical constraints ease. Amazon removed its token leaderboard, Microsoft canceled code subscriptions, and there have been reports of unexpectedly large token bills. The salient point is that even powerful technologies must pass through cost curves and capacity constraints.
Would have expected that from an AI thought leader, podcaster, or Twitter person.
But it's funny coming from the AWS cloud gold check.
I think many would write that more AI generated code doesn't necessarily make your team faster. But it's factually incorrect for a lot of teams that are moving faster than ever.
Also, read the next post in this thread; it's clearly AI generated.
The real bottleneck was never writing code.
It's releasing it, debugging it, keeping it running well. When Honeycomb CTO Charity Majors set a productivity target, she chose 2x and built from there.
You think this is AI? The syntax feels AI.
I think it was human edited. Her team skipped mandates and built AI values instead. Every AI output has to have a human owner.
If you don't want your name on it, it's probably not good work. Quality first, quantity second. Coming from AWS, that's a wild statement, especially with the backdrop of a $500 million budget. And Amazon is one of the largest investors in OpenAI and a leading coding model provider. They didn't say more AI generated posts don't make your social media team faster. They got 14,000 likes on that post.
Shriram Krishnan, an investor and advisor, said not many VCs took Bending Spoons' intro in 2022, but from 2024 onward many clamored for an introduction. Bending Spoons is going public. They own AOL, Eventbrite, and Vimeo, and maybe Evernote. Their employee growth chart spiked after COVID. They are known for operational excellence. There's a time for aggressive expansion and a time to consolidate.
Why is this guy so obsessed with GTA?
John, have you ever played GTA? Any GTA?
Maybe GTA 3, GTA 4 for like two minutes.
Two minutes? What a loser. Never gamed in his life.
I couldn't get into it. Even as a child, I thought everything you can do in the game you can do in real life minus.
No, you can't. It's all crimes.
Jordy turned it down. They wanted to pay me to play GTA, but I turned it down.
They wanted to pay you to play.
One of my first jobs was a video game tester. I got paid to play video games. I was already hooked. You play games while in development, file bug reports. It's quality assurance. I worked with a guy who made it his career. He later became a DVD tester, watching full movies to check for bugs. That was a wild job.
This fellow is obsessed with GTA 6 and is doing anything to understand if it will be released on time. He set up acoustics around Rockstar headquarters to measure audio levels, camped outside monitoring traffic, and even measured oxygen levels to infer how many people were working. He noticed oxygen levels dropped after 5 PM, suggesting increased consumption. He also monitors cigarette butts. Maybe he's a shareholder.
I don't believe people go to these lengths unless they're making money on it.
Okay. And how would he make money?
There are prediction markets with volume, or trading the stock directly.
Yeah, absolutely. Fun story. A hedge fund should hire this guy.
It almost feels like it's going over the line, maybe illegal. I'd be frustrated if someone was outside our Ultra Dome measuring everything to see if we're going to interview Matthew Prince next.
And we are because he's in the waiting room. We're bringing him into the TBPN Ultra Dome. Welcome to the show, Matthew. Thank you so much for taking the time. How are you doing?
Good to see you. I'm doing well.
What's new in your world? You made an acquisition. Take us through it.
We bought a company called Void Zero, which makes Vit, one of the most popular developer platforms. Incredible team. Evan Yu, the founder, is a first-class human being. The team is great. Vit is increasingly powering agents on the internet, and many of those agents run on Cloudflare, so it's a natural combination.
How simple is the synergy? Will you funnel those 130 million users who download Vit monthly into preferring Cloudflare, or is there more synergy under the hood?
We plan to continue it as an open source project and invest in it. We want to integrate it closely with Cloudflare's developer platform so that Cloudflare is the best place to run Vit projects. But it'll work on other platforms too. We wanted Evan and his team to have support to invest in the platform, and that will drive more developers to Cloudflare's Workers platform.
What are the headaches for developers these days? Token costs, uptime? What's at the top of the stack and how are you helping?
I think you need a different architecture for what's coming with agents. There are about 100 million knowledge workers in the US. If each had one agent running in a traditional container, that would require about 50% of total CPU capacity. For the world, it's 30-40 times current capacity. Cloudflare Workers is built on isolates, not containers, which is much more efficient. So agents are a natural fit. That's why big AI labs like OpenAI target Cloudflare. With Vit as a first-class citizen, we can help power that future.
Can you help me understand CPU bottlenecks and solutions? I'm thinking about bandwidth, CPU load, and maybe conventions like robots.txt or MCP servers delivering lighter JSON packages. Is there a path to optimization? Are we screwed or is there an opportunity?
Two different problems. First, when AI systems perform complex tasks like planning a vacation, they need coordination and code execution. LLMs are good at writing code, and that code needs a place to run. Containers bring in an OS and tooling, making it heavy. Isolates are much lighter, so more agents can run on the same CPU. Second, for agents consuming the web, we automatically convert content to markdown, which saves tokens and processing. So we're optimizing both on the developer side and the content side.
John Gruber total victory. He invented Markdown.
It's ahead of its time. It's key for making information as compact as possible.
John Gruber created the format for God. Funny world. Talk about last week you announced that the threshold had been passed around agent versus human traffic. Did it happen sooner or later than you expected?
Much sooner. At the end of 2025, I said bot traffic would pass human traffic by the end of 2027. About 3 months ago, I revised to first half of 2027. It actually happened in the first half of 2026. It shows how quickly this is growing. Agents have boundless attention; they might visit 5,000 websites for research. That drives enormous consumption. Also, the web has started growing again exponentially after plateauing since 2015. In the last 18 months, we're back to early 2000s growth. I wouldn't be surprised if in 5 years, bot traffic is a thousand times human traffic. We've got to make the internet work for that future.
Yeah, every year we'll add another nine to the percentage of bot traffic. Do you know the baseline pre-AI?
It was about 20%. Google was the largest, plus malicious bots. It was stable since 2010. Then it started growing steadily, accelerating in the first half of 2026.
Talk about the inference stack. We see a bifurcation: Apple on-device inference, NVL72 models, Cerebras chip. How does Cloudflare fit in? Will you offer inference at the edge?
It's funny. In 2022, we announced a partnership with Nvidia to put GPUs at the edge. At the time, nobody cared. Two years later, everyone cared. We basically reissued the press release. Today you can run inference at the edge of Cloudflare. We're in over 350 cities, within milliseconds of most of the world's population. My assumption is that 50% of inference will be on device, but there needs to be a protocol to hand off longer-running tasks to the edge, and then to centralized data centers if needed. Actually, I think less will be on device because tasks are becoming longer-running, like planning a vacation that might take minutes or hours.
What are the decisions when doing inference at the edge? Shaving off milliseconds may not matter for a 20-minute workflow. Are you optimistic about a state where inference needs to happen within the request loop?
Three buckets. First, latency for human-computer interaction. Second, privacy and regulatory reasons, especially in Europe wanting data sovereignty. Third, we can be more cost effective because bandwidth is effectively free for us, and we often don't pay for space or power, passing savings to customers.
There's debate over data centers: how big they should be, regulation. How does that affect Cloudflare? How are you positioning your footprint?
We'll need more data centers. Some concerns are silly, like water usage; golf courses use more. But real concerns about power. Cloudflare is different: we spread machines globally, often in the oldest data centers where networks meet. We have a ton of machines but spread out. We're less impacted by new builds; we need to get into existing data centers and be power efficient because we're guests in those spaces.
Italy and Spain have been puzzling. Football leagues there have been aggressive against me and Cloudflare over piracy. We don't like piracy either; it costs us bandwidth. We work with other leagues like NBA, NFL, MLB to take down pirates. But La Liga and the Italian league have chosen to sue us instead. It cramps my summer plans.
I want to talk about IPOs. Life as a public company. Your stock is up over 100,000 basis points. Advice for entrepreneurs trying to go public?
I love being a public company. Private markets are like a jurisdiction without no-fault divorce; it's hard to fire VCs and they can't fire you, so it's dysfunctional. Public market investors are more honest; if I do something stupid, they sell and we have a real conversation. That's felt much more honest.
Healthier than what you see in the market.
It's so true. I mean, there's every angel investor in any number of meaningful companies is going to have portfolio companies where you've fully given up on the company and you just try to be nice and help if you can, but at some point you're just totally disassociated with the investment and you're just like, and that's a reality. But it's very healthy to be like, yep, I'm disassociated, moving on, put my energy elsewhere.
In both directions, right? It's just a better relationship.
I loved the process of going public. I think the thing that we did, it was an opportunity for us to really kind of retell our story. You don't get to reinvent yourself very often, but the process of writing the S1 was an opportunity to sit down and really do it. We dug in and told the story in a way that to this day we still refer to parts of that document to explain what is Cloudflare and how did we work. I think the most clever thing that we did, and this was advice that I got from Ryan Smith from Qualtrics, was he said, 'What are you doing about friends and family?' And I'm like, 'We're not going to do it because you can take 5% of an IPO and allocate it to friends and family, and I didn't want to get my aunt upset if the stock went down or something. I didn't want to have to explain that over Thanksgiving dinner.' And he said, 'No, no, no, you're thinking about it wrong.'
Think about the people who, if they owed you a favor, that they could make a meaningful difference in the future of Cloudflare and then offer them the ability to invest in the IPO.
And I was like, 'Is that legal?' He's like, 'Check with your lawyers, check with the bankers, but the answer is yes.' And I was like, 'Some people are going to have conflicts.' He said, 'It doesn't matter. Even the fact you offered it, even if they can't do it, will mean that they'll always remember that super fondly.' And it was incredibly good advice. We made this crazy list of all of these people who might be an important relationship for us. And like 75% of them said yes. And they all made a ton of money as a result of it. And it was one of those kind of...
Now your aunt doesn't talk to you anymore.
Like I missed out on a 100,000 basis point move. How could you?
I am always comforted that we actually priced it at $15 a share and it went up to... That's actually the other thing that's interesting. Everyone, Bill Gurley and I have fights about this from time to time about IPOs. You have a lot of control over how much the pop is going to be. We sat with bankers and we were like, 'Listen, we want it to go up about 20%.' And they were like, 'Okay, if you want it to go up 20%, you price it right here.' So we priced it right at that point and it closed the first day at 18, which is exactly up 20%.
Okay. But just to steal a little bit, is that a function of your business? Not to be rude or anything, but there are some businesses that are just like meme stocks out of the gate and they're in some weird thing and there's some froth and they have less control, or is it always controllable? And part of that was, did you think the fair value of your business was 18 or 20% higher than where you priced it? So you wanted the market to sort of reprice and you didn't care about not...
Yeah. I mean, if we had priced it at 13, it would have gone up more.
