Anthropic Investor Franklin: AI Safety Concerns Won't Slow Spending

0:00 Friday. Net net, where have we ended this week? Nice ending to, what Shireen just talked about, giving us data about the 26%, r help from, agents in the r and d work. I think that's a good ending versus the started at the beginning of the week. Yeah. I'm definitely in the anti, camp. I think that with I'm I'm I'm appreciative that these founders are talking about guardrails. They're talking about making sure that everything is is safely done. But I also think, you know, we shouldn't be slowing down our pace here. And even if we did on the pretraining side or the training side where there's a lot of expenses on that front, inference is going to continue. And that means the spend will continue. And as a function of that, you're gonna see, you know, a number of companies that pulled back, you know, stocks in the equity market pulled back on Monday with fears of, oh, slow down spending. That spending can't slow, especially This is why you I I am grateful to have you here. You are a public market investor, and you are a private market investor. And so, like, let's go to public markets for a second. The idea was if the labs are pacing at the frontier and that pretraining or training work is impacted, the question posed, does that mean a pullback on the spend Exactly. For the compute? The net conclusion on that is No. Because inference need requires a lot of compute. You've seen a number of deals that these companies have done the last few months to secure compute deals. Do you think they would have done that if they thought they were gonna pull back if pacing was gonna pull back spend? We don't think that would have happened. And then it's interesting because, yes, that narrative was being talked about on Monday. I was at an industrials company on conference on Tuesday. And at that conference, a number of these infrastructure companies that are exposed to data center were saying, our businesses are diversified. We have non data center exposure as well because they saw the reaction in the market. But on Wednesday, some of that had already cleared up. Both businesses are going to continue to grow.

1:57 It's really hard to frame. There are lots of different camps on the right course of action. All of them accept that the capabilities of the models are improving and outpacing the rate at which alignment and safety keep up. Right? That that seems to be Yeah. A point of agreement. What Dario Amodei advocated for in his essay on Saturday was independent evaluation, but also the idea that that antitrust waivers are appropriate so that that collection of Frontier Labs can work together. How do you feel about that?

2:27 Yeah. Look. I don't know if they really want widespread spread regulation and federal reg we've already seen You don't think I I don't think that. They I do think that they want, good actors here, and they every company to put guardrails in place and make sure that everything is done in a safe way in a sandbox. And we've seen examples of that that that hasn't happened, unfortunately. Which is the school of thought. Like, you're the frontier labs. You're the experts. You should have a handle on this, and don't release the model if it's not safe.

2:56 Exactly. On top of that, though, some are talking about the fact that there are a number of security products that that are gonna be launched very soon, and so we can also protect you. You're you're you're gonna see some of those product launches come. Again, couple of weeks ago, a tech conference, the entire theme at that tech conference was cybersecurity. Right. Whether it's the public companies exposed to this or the, venture community was talking about what some of the frontier labs are gonna start to offer and a number of private companies You help us understand this. You I think you heard Ali, the Databrick CEO on the show yesterday. Like, he went to cybersecurity straight away. Andrew O'Day also brought it up.

3:32 Loads people brought it up. The idea is that it means very strong cybersecurity platforms are more important than ever. They're incredibly important. Not only are they important, but also just pacing these models. You haven't seen widespread adoption of all these models right now and outside of, Anthropic talking about it right here, right now. Right? Yes. So so corporations are gonna start you know, you have GPT five, GPT six, you know, Fable coming to market. Companies haven't acknowledged workers are not even used to using these right now. You just with the launch of Muse, Instinct, some of these agents that consumers are starting to work with, I think that will bring some adoption on the consumer side, and consumers are gonna start asking for that in the workplace just like you saw with the iPhone. Years you know, it took a while for consumers to move from BlackBerrys to iPhones. Right? But at the end of the day, it was they started seeing what it benefit what kind of benefit they get from this, and then they started asking it for asking for it in their workplace.

4:30 Sara, I I appreciate there's a limit to what you can say as an existing investor in Anthropic. But a a question posed by the market is, are these safety concerns the real reason why OpenAI will not do an IPO until 2027, which Sam Altman said in an interview. Actually, there was plenty of reporting that the IPO would be 2027 anyway. But are the safety concerns a real reason for the market to slow down and say, like, we need to to stop funding this in private or public markets?

5:01 Again, I don't think so. If if in you know, we're seeing we're starting to see some of these benefits, the productivity gains from these models, the benefits that society will see, not the 10% or wherever that number comes from in terms of, humorism. The economic upside. The the upside that's gonna we're starting to see that diffuse across a variety of industries. You've seen that the market start to broaden as a result of that. So we we shouldn't stop funding. And I think, you know, IPO timelines, I think some of them were already being pushed out to 2027, and I think we're gonna start to see some of these come through.