transcribe

Anthropic Shares Fall After IPO, Nvidia’s Future and Bending Spoons’ $3.68B Mailchimp Deal

The Information · 27m · transcribed 3h ago
More from The Information Business
𝕏 Share ▶ YouTube 📥 PDF 🤖 .md

Section Insights

# 0:00

Introduction to Market Trends

What are the recent developments in the stock market related to Anthropic?

Anthropic went public recently, with its shares initially trading above the IPO price but experiencing a decline shortly after. The market sentiment appears fragile, particularly for AI stocks.

  • Anthropic's shares opened lower than expected after going public.
  • Initial trading showed volatility, with significant fluctuations in stock price.
  • Market sentiment towards AI stocks is currently negative.
# 5:27

Investor Sentiment and Stock Performance

What factors are influencing the stock performance of Nvidia?

Nvidia's stock is under pressure due to concerns about the AI market's growth and uncertainty regarding its leadership and future revenues, leading to significant sell-offs.

  • Nvidia's stock has seen a sharp decline over the past month.
  • Investor uncertainty about AI growth is impacting Nvidia's stock negatively.
  • Leadership changes and market inquiries add to the volatility.
# 10:55

Bending Spoons Acquires Mailchimp

What is the significance of Bending Spoons acquiring Mailchimp?

Bending Spoons is acquiring Mailchimp for $3.68 billion, indicating its strategy to consolidate older internet brands and revitalize them for growth.

  • Bending Spoons is actively acquiring established brands to create a growth story.
  • Mailchimp's acquisition reflects a trend of tech companies seeking to innovate within legacy brands.
  • Bending Spoons aims to streamline operations and improve tech stacks post-acquisition.
# 16:23

Operational Strategies of Bending Spoons

How does Bending Spoons plan to integrate its acquisitions?

Bending Spoons intends to enhance its acquisitions by implementing operational efficiencies and sending its own employees to improve tech and workflows.

  • Bending Spoons focuses on making acquired companies more efficient.
  • They plan to send employees to integrate and improve tech stacks.
  • The company has a history of increasing prices post-acquisition, though not for all brands.
# 21:51

Harmonic's Role in AI Safety

What is Harmonic's approach to ensuring the safety of AI-generated code?

Harmonic focuses on verifying software and hardware to prevent vulnerabilities, particularly those arising from legacy code, and emphasizes the importance of formal verification.

  • Harmonic aims to address cybersecurity concerns related to AI-generated code.
  • Most vulnerabilities tracked come from legacy code rather than new AI-generated code.
  • Formal verification can significantly reduce memory safety vulnerabilities.

Transcript

0:22 Heat. Heat. Heat. Heat. >> >> Welcome everyone to the information ti.

1:16 My name is Akos Pasicha. Happy Friday to you. It is November 20th. It's been a busy week, but we are almost through to the weekend. Anthropic went public yesterday. We are going to go through all the numbers around how it traded with you today. Shares fell this morning. It is still above the IPO price, but we're going to bring on Martin Pierce to break down all of it with you. We are are also going to talk about bending spoons with our finance reporter. The company said it is buying Mailchimp for more than $3.6 billion.

1:51 We're going to talk about how the company is financing these types of deals. And to close out the show, we are going to talk to the CEO of Harmonic, the AI company co-founded by Vlad Tenev. It's going to be a great show. So, let's get right on into it. Anthropic shares opened just shy of $90 a share this morning, which was more than 7% lower than it closed last night. Yesterday, of course, the stock popped from its $83 IPO price to more than 115, but it spent a lot of its first day as a public company in the high $90 range. I want to bring on Martin Piers, our editor at large for this week's edition of The Editor's Cut to talk through all of this. Martin, welcome back to the show. It was great to have you here.

2:36 >> Good morning, Akash. >> What do you want to talk about? >> So, right now, Anthropic stock is trading around 8691 down 11%. Looks to me like the employees are able to unload 10% of their stake are doing that. which is probably depressing the price. this is not a great sign. It does suggest that demand for anthropic is fragile, but it also suggests that the negative feelings we've had about the AI stocks is really here to stay. so I think we are in that period now where everybody has gone off AI a little bit. It's probably going to be harder for other companies to go public. yeah, it's an interesting time. Yeah, I mean, look, I was SpaceX is certainly the most recent proxy, but you know, SpaceX, it was kind of a weird IPO. It was a one-off. I was sort of thinking back to like the Circle IPO and the Figma IPOs, which had massive pops initially, and there was like it didn't seem like there was as much euphoria yesterday as there were during some of those listings, at least for >> That's definitely true. And I am embarrassed cuz I thought there would be more euphoria and I had predicted in my column the stock would get to 150. I was completely wrong once again. but as one of the guests on our live video discussion yesterday made the point that the market cap of these companies SpaceX and and andropic is so big that getting a a pop of you know like a 50% or more is just so much harder and also the valuation relative to the actual business in both cases is very steep.

4:30 So it's actually it's a you know it's a good sign that investors are not that crazy to be bidding up the stock that much. >> Right. And Martin can you just remind us you mentioned the employees selling the shares. just remind us of the lockup period cadence for here for Anthropic because one of the points that Jay Ritter made yesterday on our live video summit was he actually looked at the trading volume of the shares for SpaceX as the lockups came through over the course of the last 6 months or whatever and and he I think what he said is that the trading volume wasn't as high as he had thought and so maybe that suggests his employees weren't selling as much as as was feared. So how should we think about this for anthropic going? Well, that's a good point. And in fact, often you see when companies have those u lockups expiring, you see the stock falling before that, which I think is often investors expecting the stock will fall or expecting the stock will come under pressure. And so they sell down beforehand and so by the time the thing actually happens the stock is down and you know it's almost like a buying opportunity at that point. But today I have to believe that there's just not enough buying demand to offset those people who are divesting their stock. I think it probably is the fact that employees are allowed to sell you know one tenth of their stake because normally you you can't sell for 6 months or so, >> right? Okay, let's let's talk about Nvidia because I'm I'm looking at shares as we're taping this. So 6% down this morning almost 14% down in the last 5 days and more than 36% down in the last month. So I mean I guess the question I have for you is like where do you think this is going to stabilize like how low does Nvidia fall?

6:35 >> I think there are two things going on. One is that what I mentioned about the AI feeling is obviously affecting in Nvidia. I mean that stock had already fallen before Jensen you know was u pushed out. That stock had already come down from its highs. it's obviously you know the one of the most exposed to the AI trade. So if you're not sure about the growth of that market, Nvidia is a company that you might be concerned about. So there is that issue.

7:06 But obviously the much bigger issue is that we just don't know what will happen to that company. We don't know who will be in charge of it over time. We don't know what the inquiries will discover. We don't know what the revenues will be like. I mean, Jansen was very good at doing all these deals to juice demand. Now, they have pulled back from those deals. It's there's a lot of things that we don't know and I think that's the reason why people are dumping the stock and there's just it's again there's just not that much buying demand. I'm sure it will stabilize. I'm not sure where but I think I looked the other day and I noticed that the stock had been around 150 in the middle of last year. So it can go down a little bit more before it gets down to, you know, really cheap.

7:59 >> right. >> Yeah. Yeah. No, I mean I mean last year Yeah. or earlier this year. No, you're right. I mean it's not it's not this is not not uncharted terrain for the stock. >> Not at all. I mean I think if you go back to 2022 before this whole AI thing began the stock was I have a feeling it was around 20. Could that be right? >> you're talking about when you when in 2022 Yeah. Yeah. 20 25 26. I mean this is >> Yeah. So, you know, maybe a 440. I think it will stabilize at some point because it'll just be a screaming buy. But we are in, as I said, this period where AI is just not the thing that people want exposure to right now.

8:45 >> Yeah. Okay. I I'm I'm going to tell you my prediction here, okay, about Nvidia and let's see if it happens. I think everyone is comparing this to Jobs, Cook, etc. I think Jensen founds another company. I think Nvidia buys the company and I think he's back within two years. I think he becomes CEO again following that acquisition. What do you think of that? >> I think it's look I that's a great prediction, Akash. But the problem is until the board inquiry and the SEC inquiry and the DOJ inquiry and all these inquiries into Nvidia's Chinese dealings get done. We won't really know what his future is. I think it's quite possible that one of those inquiries will really not be good for him and he won't be able to come back.

9:43 If he is cleared of the worst of these things and the worst thing you can say about him is that he had these affairs. then maybe if if Nvidia stumbles next year or the year after that then maybe they bring him back. But we're also seeing some very negative stories. is there was one in the New York Times this morning about how he would routinely really treat people badly and had done so for years. So I don't know whether his reputation now is really taking a big hit and I I'm not sure that it will actually survive that to the extent that he's able to return. But it really does also depend on what these various inquiries find and how the business actually performs. Right. Okay. Well, we'll revisit in two years. Martin, I want to thank you for coming on the show as usual.

10:42 >> Years. Who will remember that we even talked about this then? >> I will. That's the point of making a prediction is that you put it out there and then if it comes true, you cut it and then you play it in two years for people. >> Okay. >> Okay. All right. Well, I want to thank you for coming on Martin. That's Martin Pierce, our editor at large here at the information. Okay, switching gears. Tech conglomerate Bending Spoons is buying Mailchimp, the email marketing company from fintech firm into it for $3.68 billion. Bending Spoons has been very active in the debt markets to finance acquisitions like these. My colleague Alex Iikinstein has new reporting on another funding vehicle the company could use soon. I want to bring her on to talk more about it.

11:28 Alex, welcome back to the show. It's great to have you here. >> Thank you. Excited to be here. >> Can you just remind us like Bending Spoons? I mean, I I hear the name everywhere and like what is it? Like it's just a bunch of companies together. Is that the idea? >> yeah. I feel like they've made a lot of news in recent years. Everyone wants to talk about them. They're a Milanbased technology holding firm that has picked up a few more established or older recognizable internet brands.

12:00 So they own AOL, Eventbrite, Vimeo, and then this year they picked up Air Table and Murro. I It's so funny cuz when you think about the that collection of names, it's like all the names that people at some point said this business is done. like you know this this is the hallmark business of the past wave of the internet. but Bending Spoon says no we can if you put them all together there's still a growth story here I guess.

12:31 >> Yeah. And I think Mailchimp is sort of part of that. >> Yeah. Okay. And and Mailchimp remind me that's like the it's like the mail merge software right like all the all the marketing emails. >> Yeah. I think they gained a lot of prominence when newsletters were becoming all the rage a couple years back and they've since struggled with growth since like 2021. They've stabilized because there's more competitors like Substack and things like that out there.

12:58 >> Right. Okay. And and so I mean we we'll talk about the Mailchip deal in a second, but how are they funding all of these acquisitions or how how have they? >> Yes. So they've done an a mix of debt and equity raises. after going public in July. they have about 5 billion in long-term debt right now and 4 billion in net debt as of the end of September. They just released their earnings call for the third quarter this morning.

13:28 >> Okay. And where does that put their their debt ratios at? Is this a high number or a low number compared to other companies? >> Pretty high. Yeah. And we can get into it, but it has affected their investment rating. Okay. What how has it affected it? >> Yeah. So, their investment grade is below below their credit rating is below investment grade. It was upgraded a little bit in October before they had done the Miro deal. after they had did a $1.8 billion high yield bond.

14:03 it means that they have to pay a little bit more in terms of borrowing costs. and while they could lower that hypothetically if they paired down their leverage and did a few other trade-offs, a source had told us they're not really interested in doing that right now. >> Okay. Now, the what you wrote about this week and and you published a story actually just before the the Mailchimp deal got announced so it was very timely actually is you wrote about how they they may be looking to explore convertible bonds as a new funding vehicle. We'll talk about the strategy here in a second, but can you just remind us how do how do convertible bonds actually work? Yeah. So, convertible bonds are loans that investors can eventually convert into shares and they're good because they come with lower or low interest rates.

14:55 but at the same time, they come at the risk of diluting existing shareholders. >> And and when do they typically convert? Like is this a is this an optional thing? Is it a time period thing? What what's the cadence? I believe it's up to the individual shareholders. >> Okay. Okay. So, now you said the the interest rate is is different here. So, Bending Spoons is going about this route. Why?

15:25 >> the potential convertible bond route, you mean? >> Yeah. Yeah. Yeah. Yeah. >> Yeah. I think it's they would only do it as a complement to their core like debt and equity raises that they've been doing in recent years. know they did two they did a Birkshshire Hathaway hack Birkshshire Hathaway equity deal a couple weeks back because it wouldn't be able to raise like as much as those debt and equity deals have but it would be like a nice complent to help them in their acquisition strategy which remains really active as evidenced by the Mailchimp deal and to run the existing companies that they already have in their portfolio.

16:03 And maybe coming back to the Mailchimp deal, I wonder in your reporting did you find sort of what the playbook has typically been for Bending Spoons in terms of what they do after these acquisitions and what that might suggest about about Mailchimp? >> Yeah, they've made a lot of noise because the companies that they've purchased they'll go in and do pretty major layoffs u to make the team leaner and increase prices for customers. they've increased prices at about more than half of the companies that they've purchased so far. Although a source told us they're not planning to do that with Air Table which was one of their more recent acquisitions. and then they will send in about 50 to 60 of their own employees. They call them spooners to increase the tech or improve the tech stack and just streamline operational workflows to make it more embedded.

16:56 >> Wow. And I'm just goo. Do you know the answer? Why why is it called bending spoons? Have we have we decoded that? >> No, I don't know what that answer is. I never bothered to ask. >> All right, let's see. Let's see. The tech company Bending Spoons got its name from a scene in the science fiction movie The Matrix. >> Got it. Okay, makes sense. >> There you go. Okay, this is where people say, "Duh."

17:22 All right. Well, I want to thank you for coming on, Alex. It was some great reporting. That is Alex Iikinstein, our finance reporter at the information. Harmonic, an AI startup co-founded by Robin Hood CEO Vlad Tenev, is looking to capitalize on growing concerns about the safety and reliability of AI generated code. The company is valued at $1.45 billion. For more on this, I want to bring on Harmonic CEO, Tutor Ake. Tutor, welcome to the show. It's great to have you here.

17:52 >> Thanks, Akash. Thanks for having me. So what does Harmonic do today? >> Yeah, Harmonic is out to verify all the world software and hardware. We've started with C, Rust, and Verarilog. We've got more languages coming soon. and I totally agree. I think that one really important weapon in the toolbox against cyber security is to verify the code and that's what we do. >> Okay. And you guys started with math, right? I mean, you started with AI models specifically for math problems.

18:20 Yeah, we thought that you know starting with a model that's really good at math, especially formally verified math would make it really smart and generally capable at other kinds of reasoning including software and hardware verification. So the core technology we built which is formally verified mathematical reasoning has now been applied to formally verified code and hardware reasoning. >> Okay. And is that how you you met Vlad? Like what's what's the backstory here? >> Very briefly. I was walking with the dish with a friend back in 2023 and I said, "I'd love to try to build an AI that could win a Fields medal." And the next day he met Vlad Tennv at the Robinet offices and Vlad told him the same thing. and then he connected us.

18:59 >> Oh, it's that easy. Okay. >> We're lucky. >> Maybe we should all just start putting out, you know, bold really, you know, awesome ambitions and then >> at the Dish. It's a great hike, Heather. >> Yeah, I know. And then you just, you know, you have like, you know, I don't know. Y just matches people with bold ambitions. that that there's a a matching service. okay, so just walk me through then the pivot here. So if you started with sort of the math focus, why have you just sort of decided on this verification AI code security product now?

19:31 >> Well, I mean to be clear, this has been a long time coming. We started working on Verilog, which is a hardware language in October of 2025. So shortly after we won the International Math Olympiad. also in the the seed deck from 2023, our vision was to verify all the world's software and hardware by 2028. So, this has been the mission all along. I think that we've just been surprised by how well the technology generalized and you know, we've been happy to be able to produce models and product that just works super well at FE.

19:58 >> Okay. And and so who are your customers now and and what what is the product you're actually selling? >> Aristotle is our product. on day one you can formally verify all of your C, verarilog and rust. a lot of the world's safety critical code is written in C. Almost all the world's hardware is written in Verarilog. Our customers tend to be large hedge funds, semiconductor companies, cyber companies. Basically, anybody who's worried about attacks, anybody who's worried about shipping software and hardware faster with AI benefits from formally verifying the code. If you think about it, correctness is kind of like the last rate limiting step to letting AI loose on code bases. And if you can formally verify code, that is just one tool in the toolbox of correctness, but it's a very very powerful one.

20:45 >> And and forgive me here because I'm not as deep in this. but when they talk about code review because you've got these AI code review platforms, is is code review and then verifying code before you deploy it like is that the same thing or are these two different tasks here? Formal verification is building a mathematical model of what code and hardware does in the real world and then proving things about it. So it can be a tool to help code review. It can be a tool to replace code review in some cases. It's just an independent way to get extremely high assurance about what the software hardware does.

21:22 >> Got it. Got it. Okay. So I mean this is a very timely business then given all the security vulnerabilities that we've been seeing. we had Black October, I think they call it. Now we have November and hasn't really stopped. So, you know, one of the questions I wanted to get your read on is we know that AI has been helping people find these vulnerabilities. I'm curious, do you have any sense of what evidence there is for AI creating any of these kinds of vulnerabilities? you know, maybe the AI generated code hasn't been reviewed or verified properly, maybe has bugs that could have otherwise been avoided.

22:06 >> So, what we see is that in the public CVE database, most of the vulnerabilities I'm tracking come from legacy code, not from new code. now, I don't know what's happening with the recent outages. It could be something different. But the important thing is that you know we wrote a lot of code in a world where AI couldn't attack it. And so maybe we weren't quite as careful about how we wrote it. we weren't quite as defensive. Now if you run a tool like formal verification your codebase it doesn't solve all the issues but it really solves a lot of them. So if you look at the top CVEes, probably 30 or 40% of the ones that are the most severe that lead to something like code remote code execution or the it comes down to something like a memory safety vulnerability. That's a kind of industry term for something where a program does something it shouldn't do with memory it has access to. And formal verification is an easy way to just completely get rid of those vulnerabilities. going forward, I think that as AI writes more code, you also want to be careful with it. But I don't really have evidence right now that AI generated code is what's leading to all these recent bones.

23:10 >> Right. Okay. tell me, you know, a business like yours, I sort of wonder how you might think about competition from the labs and if the labs were to just create their own competitive offering. what what would be your defense there do you think? >> I think all the companies are looking into how to apply formal verification. the reality is this is like an enormous market. I think we're first with the product and it's a great one and people are adopting it. we also integrate you know model development product engineering and so we make it really easy to use. I think in a lot of cases formal verification has been kind of a research project but today with a product Aristotle it's something you can put in a developer's hands and have them get value from day one. So I think you know longterm we'll see what happens but you know we're feeling pretty good about our position.

23:58 >> Have you gotten any acquisition offers at all? >> I think people reach out all the time. I think you know we're not interested. >> The reason I'm ask I'll tell you the reason I'm asking is because we had Mel Lutzky from Graphite on I think over a year ago when we started the show and I asked him the question. I said you know you're doing AI code review. What's to stop cursor from from from doing something similar? And he he made the point he said look sometimes it's helpful to have a third party doing the reviewing not having the coding company doing the reviewing. And then graph cursor went and bought graphite. So I don't know may maybe that's a future in your cards.

24:39 >> look we've had we're very focused on it. I >> you don't have to comment on that. It's fine. It's fine. I'm just I'm just sharing the story. It's it's sort of the way it goes. okay before we we go though so given that you have your your roots in in math you know we had this anthropic mathematician claim today that they have discovered a proof of the Raymond hypothesis can you just maybe help us understand the significance of that and and you know what this could mean for these models going forward?

25:16 Yeah, I think I think that it's so Levven is a highly qualified mathematician. I definitely believe that he has a proof. Also, he released the lean artifact. That's the same language we use for theorem proving for our programs. I think this is a testament to how well, you know, asynchronous reinforcement has worked for the past 2 years to just make models really smart. I don't think we should be surprised at this problem being solved, at future problems being solved. I I think AI will just continue to get smarter and smarter at many of the world's challenges. and it's obviously amazing work and we should expect to see more of it.

25:47 >> Okay. And last question for you. I mean, you know, AI being able to solve all these these really difficult math problems and yet we still have some basic issues with, you know, it completing you know, everyday tasks consistently, for example, the agents and stuff like that. Why why do we have this this disconnect right now? Yeah, I think it's a question of, you know, how well you're able to apply that tool. I mentioned reinforcement learning. So, in a domain like math, which is super verifiable, it's really easy to scale up. in a tool like, you know, writing emails or having the right tone when communicating with people or robots in the real world, it's a little harder to get the perfect reward signals. And as a result, I think that some of these fields trail behind others. But I think you know the the bulk of the evidence is that AI will keep getting better at most tasks, some faster than others as a function of how verifiable they are.

26:42 >> Great. Okay. Well, Tutor, I want to thank you for coming on. That is Tutor Keem, the CEO of Harmonic here on TIV. >> Thank that does it for today's show. A reminder, we are on this stream Monday through Friday at 10 a.m. Pacific, 1 p.m. Eastern. If you can't make it, then episodes are available on the information.com, on our YouTube channel, or wherever you get your podcasts. Make sure to follow us on social media, on X, on Instagram, on Tik Tok, and on LinkedIn. I am already excited for our next show on Monday. Have a great Friday. Have a great weekend. Get some rest. Bye-bye for now.

Summary

Anthropic's recent IPO saw its stock price fluctuate significantly, reflecting a cautious market sentiment towards AI stocks. Bending Spoons is acquiring Mailchimp for $3.68 billion, utilizing a mix of debt and equity financing, while Harmonic, co-founded by Robinhood's CEO, aims to enhance software security through formal verification of code.

- Anthropic's stock opened at $90, down from its IPO price, indicating fragile demand for AI stocks.
- Employees are selling shares, contributing to the stock's decline and reflecting negative market sentiment.
- Bending Spoons, a tech holding firm, is acquiring Mailchimp to expand its portfolio, which includes several established internet brands.
- The company is financing acquisitions through a combination of debt and equity, currently holding significant long-term debt.
- Harmonic focuses on verifying software and hardware security, leveraging its expertise in formal verification.
- The startup's product, Aristotle, aims to ensure the correctness of critical code, appealing to sectors concerned about cybersecurity.
- The conversation highlights the challenges AI faces in consistently performing everyday tasks despite advancements in complex problem-solving.

Questions Answered

What are the recent developments in the stock market related to Anthropic?

Anthropic went public recently, with its shares initially trading above the IPO price but experiencing a decline shortly after. The market sentiment appears fragile, particularly for AI stocks.

What factors are influencing the stock performance of Nvidia?

Nvidia's stock is under pressure due to concerns about the AI market's growth and uncertainty regarding its leadership and future revenues, leading to significant sell-offs.

What is the significance of Bending Spoons acquiring Mailchimp?

Bending Spoons is acquiring Mailchimp for $3.68 billion, indicating its strategy to consolidate older internet brands and revitalize them for growth.

How does Bending Spoons plan to integrate its acquisitions?

Bending Spoons intends to enhance its acquisitions by implementing operational efficiencies and sending its own employees to improve tech and workflows.

What is Harmonic's approach to ensuring the safety of AI-generated code?

Harmonic focuses on verifying software and hardware to prevent vulnerabilities, particularly those arising from legacy code, and emphasizes the importance of formal verification.

© transcribe · For agents Built with care and craft by Gokul Rajaram