- 9 minutes ago
Smartbird CEO Nadia Carlsten is driving one of the boldest moves out of the direct-to-consumer retail era: a pivot from selling footwear to AI infrastructure.
As part of Business Insider’s The Long Play, Carlsten discussed why the company formerly known as Allbirds is now betting its future on AI and rebuilding its team.
“I think every company needs to reinvent itself every now and again,” Carlsten told Ben Bergman, Business Insider’s chief correspondent covering venture capital and startups.
She also shared how Smartbird is convincing customers and investors that its shift isn’t a gimmick.
As part of Business Insider’s The Long Play, Carlsten discussed why the company formerly known as Allbirds is now betting its future on AI and rebuilding its team.
“I think every company needs to reinvent itself every now and again,” Carlsten told Ben Bergman, Business Insider’s chief correspondent covering venture capital and startups.
She also shared how Smartbird is convincing customers and investors that its shift isn’t a gimmick.
Category
🎈
FunTranscript
00:00Thanks so much for coming in.
00:02I don't know if you guys noticed, but do you know which shoes I'm wearing?
00:06Yes.
00:07So how many of you wore Allbirds back in the day?
00:12And how many of you still wear Allbirds?
00:15One person.
00:16Two.
00:17Okay, good.
00:17All right.
00:18Well, and I know you told me you actually never in your life wore Allbirds.
00:23Never.
00:23Okay.
00:24Never.
00:25I can see that.
00:25Clearly not Allbirds.
00:28I'm not surprised, and I know you are probably a little tired of talking about this, but
00:35I do have to ask, because everyone asks the same thing, why?
00:39Why go from a sneaker company to an AI infrastructure company?
00:43Is it just because it's 2026 and everything has to be AI?
00:47Yeah.
00:47So it's funny because everybody does ask that, and kind of my question is why not?
00:54You know, why not go into AI?
00:58All right.
00:59So the short answer is the shoe business wasn't doing that well.
01:03This was before I joined the company, and they wanted to do something new, and AI is exciting.
01:10That's something that there's a lot of value in the markets for AI, and the goal of the
01:15company was to do something valuable and meaningful.
01:18For me, I knew I wanted to be an AI infrastructure.
01:22I was going to do that no matter what, no matter what the format was, and it wasn't that important
01:27to me how you do that, whether it's taking over Allbirds and turning that into an AI infrastructure
01:32company, whether it's starting a new company, like the how matters less than what you do.
01:38So for me, it wasn't as crazy as it seems for a lot of people.
01:42So maybe Business Insider could be an AI infrastructure company next year.
01:47Don't get any ideas.
01:48Only if you're actually committed to it, and you know how hard it is to be an AI infrastructure
01:53company, and you actually take the steps to do it.
01:56And we're going to talk about that, but I want to show the market was very excited about
02:00this.
02:00I want to show the chart when this was announced, the day they announced the pivot.
02:04So you got about, at one time, an 800% pop on this news.
02:10Markets were very excited.
02:12You've been in this space a long time.
02:14What made you believe this specific opportunity was worth tackling?
02:19Yeah, so I've been in computing for a long time.
02:22And for me, what's really interesting is when computing in all its forms, right, then we're
02:27going to talk about AI, but I've done cloud, I've done quantum, I've done, you know, very
02:33old school, high performance computing deployments.
02:36And what's interesting is how does that support innovation?
02:40And we're at a great time in the industry where computing is getting better.
02:44There are more forms of computing, but also the adoption of that computing is getting a
02:49lot more interesting.
02:50It's going from just a few people in labs, a few people developing models, to just normal
02:56people in enterprises, inside companies trying to do something innovative and using that
03:02compute to do that.
03:03So for me, it was clear that this is actually a good time to get into this.
03:07And all birds was the right place at the right time.
03:10And it's been five months, roughly, since you took over.
03:14Yeah, I think even less than that.
03:15Time flies.
03:16Yeah, it started in June.
03:17And how has it gone?
03:19It's been amazing.
03:20I mean, what a fun ride.
03:23Definitely never in my life thought that I would be the CEO of a former shoe company.
03:28And managing what is essentially a startup and with kind of a full build out of a new
03:34team, new technologies that we're looking into, new pipeline of customers, all of that,
03:42while at the same time being very much in the public eye.
03:44Why?
03:44Because the thing about all birds, it's a brand that is loved by so many people.
03:49So people really care how we do as a company.
03:51Except you.
03:52I don't have anything against the shoes.
03:54They seem great.
03:56It's just not my thing.
03:57I'm good at buying shoes.
03:58I can't be the CEO of a shoe company.
04:01That's good.
04:01I can be a great customer.
04:03But you are the CEO of a public company now.
04:05Yeah.
04:06And I want to show another chart, which is the performance of the stock since this.
04:11And it's gone down quite a bit.
04:14Is this something that you look at?
04:17Do you wake up and look at this chart every day and say, why did I take this job?
04:21No.
04:21I don't.
04:24Because I didn't take the job thinking this is great because there's so much excitement
04:30about AI.
04:31I took the job because I know that there's a lot more people that will be using it.
04:36AI for the foreseeable future.
04:38So the stock market is going to do what the stock market does.
04:41And I'm not an expert in the stock market.
04:44I'm probably the, I was telling you, I'm probably the CEO of a publicly listed company
04:49that cares the least about the stock market and what it does.
04:52Well, don't tell that to your investors.
04:53They know.
04:54They don't.
04:55Okay.
04:56But what I care about is making sure that we're putting the company in the right path.
05:01And that we're able to build it quickly enough and well enough that we have a sustainable
05:06business that investors will love.
05:09So what happens in the short term?
05:11Not that important.
05:12What happens in the long term?
05:13Very important.
05:14So we've been very focused on the long term.
05:16So no, I don't look at the stock market.
05:18I'm usually looking at my email and text messages from my employees and everybody on what we're
05:24doing with our potential customers and our deployments.
05:26But why then do it as a public company?
05:30Again, why not?
05:31Well, why not?
05:32Because it's down 800%.
05:34So what?
05:35It will come back up.
05:36Okay.
05:36But you could, I mean, there's, you know, I cover startups, private companies.
05:41Yeah.
05:41It seems a much easier road to go down.
05:44This seems challenging to be public.
05:46It is a much easier road to go down when you can basically hide from the world and raise
05:52a lot of money and do what you think you should be doing and then kind of come out with
05:56a
05:57product.
05:57It's also a lot less interesting.
06:00And you don't get to work with customers right away, which are the two things that I
06:04wanted to do.
06:05So yeah, I didn't go into it because it was easy.
06:07I definitely had many more easier job opportunities that I could have taken.
06:12But I do think it's been super meaningful in ways that just building privately in isolation
06:21but not being in that public eye, not being able to talk to customers, not being able
06:25to have this conversation about why enterprises need AI would have been less exciting.
06:31Yeah.
06:31Okay.
06:31Let's take this chart off.
06:32I told them that you don't have to leave that chart up there.
06:35Look, we don't care about the stock price.
06:37But I mean, you have about $200 million on your books.
06:43Less of that substantially is actually in cash.
06:46You're competing against companies like AWS, where you used to work, Google, all the hyperscalers,
06:53plus CoreWeave, who has about $5 billion in cash.
06:57So, I mean, $200 million is not a lot for AI infrastructure.
07:02You know, that covers like the furniture, but not this massive buildout that is required.
07:07So, how do you possibly compete against companies of that scale?
07:12Yeah, we don't.
07:13We have a completely different business than AWS and even CoreWeave.
07:18So, the similarities are that we're in the same business, right?
07:22We're all building infrastructure.
07:24A lot of the names you mentioned, building AI infrastructure more specifically.
07:28But the way we go about that and the types of customers and the types of deployments that we do
07:33are very different.
07:34So, for us, it's not about scale at all.
07:37And with AWS and CoreWeave, it's all about scale, typically for those hyperscalers, hence the term hyperscaler.
07:43For us, it's about the right infrastructure for the right company at the right time.
07:49And so, for us, what that means is typically we approach or a customer approaches us.
07:57They tell us a little bit what they're trying to do with AI.
07:59We design the infrastructure based on that and then we create a project.
08:04We build what they need to do and it's financed based on that project and so on.
08:08What we don't do is what those hyperscalers that you mentioned typically do, which is invest in massive infrastructure ahead
08:16of time.
08:16And then you go find a customer to take a slice of that infrastructure and that's how you make your
08:22money.
08:22So, it's a completely different business model.
08:24And our business model allows us to work more closely with customers, but it also is a lot less capital
08:30intensive.
08:31And the capital that we spend and infrastructure is expensive.
08:35So, it's not that we don't spend money, but the way we spend money is after we have a customer
08:40contract.
08:40And those sizes are also smaller based on the nature of the customers.
08:45And why would customers come to you then?
08:47I know I read your shareholder letter, so at least you care enough about them to write that.
08:51But you said that you were targeting companies that haven't really used AI before, medium-sized companies.
08:59So, where are you finding the most success there?
09:02Yeah.
09:03So, first of all, we do care about the shareholders.
09:05We don't care about the stock price.
09:07It's very different things.
09:12Where are the weird inflection points in AI?
09:15So, for the last few years, when people talk about AI and when they talk about AI infrastructure, they're all
09:22really talking about model development and infrastructure that is supporting that model development.
09:28So, huge sizes of GPUs supporting training of these models that all of you have probably heard of and use
09:36every day.
09:37What's happening is more and more, I would say, normal companies, right?
09:41Shoe companies, car companies, pharmaceutical companies are adopting AI to do what it is that they do as their business.
09:51Most of them have been in this kind of pilot experimentation stage, right?
09:55They weren't doing anything super meaningful.
09:58That is changing.
09:59So, now they have two choices.
10:00They either start doing some of this infrastructure themselves to be able to do the AI that they need to
10:05do or they can outsource that to somebody who does that every day.
10:10And that is us.
10:11So, that's basically the pitch for a lot of our enterprise customers.
10:17We understand that you're trying to do more with AI.
10:19In order to do that more, you need infrastructure that's actually built for you, that you have more control over,
10:26that you have better economics with.
10:27And it doesn't make sense for you as a shoe company, pharma company, car company to be doing that job
10:33on top of what you already need to do with AI.
10:35So, we'll do the infrastructure part.
10:37We'll let you focus on the actual business.
10:39And you have had a lot of very impressive roles, more on the technical side.
10:44This is the first time you've been a CEO, right?
10:47And also a public company CEO.
10:49So, first time as public company CEO.
10:51This is my second time as a CEO of an in-fart company.
10:55And so, what has been the biggest challenge of that?
10:57And what have you learned?
10:59Journalists showing me the stock price.
11:01Well, okay.
11:02Touche.
11:04You're going to get a lot more of that.
11:06Yeah, yeah, yeah.
11:06But that really has been the most challenging part, is the fact that this is a very public reinvention of
11:13the company.
11:13And people are watching.
11:16And I have to say, and I'm very happy for that, a lot of them are very supportive.
11:22And I do hear from shareholders, people who just pay attention, saying, you know, this is really cool.
11:28Can't wait to see what happens.
11:30So, that's nice.
11:31So, it's overall very positive.
11:33But other than that, it's very similar to my last time as a company CEO.
11:38It's a lot of focus on building the team, going after customers.
11:43The hard part is always figuring out which opportunities to say no to.
11:47There's always some projects that sound really exciting, but then you have to balance it with, you know, your resources
11:53at the moment.
11:54And, you know, is that the right opportunity?
11:56That tends to be the most difficult.
11:57So, that part is all similar.
12:00It's really just kind of this public interest in the pivot.
12:06That's been what's different.
12:07And just zooming out to the industry as a whole, which you have a really interesting perspective on, do you
12:13think that we are in an AI bubble right now?
12:17I think it depends how you define bubble.
12:20I think if you think about it in terms of are there companies that should not exist and will not
12:27exist in a few months because they went into this just really excited about AI and not having the fundamentals?
12:33Yeah, probably.
12:33I mean, every time there's a transformational technology, this happens.
12:38There's always a lot of excitement.
12:39A lot of things get funded that shouldn't get funded.
12:43But I think where I would caution everybody that worries about the bubbles is just even if we assume that
12:51there is a bubble, that doesn't mean that the technology is not good.
12:55It also doesn't mean that the adoption of the technology is going to go away.
12:59So for me, I care less about is there a bubble, is there going to be some realignment of stock
13:04prices and so on at any moment in time?
13:06And more about is the fundamental assumption that we're making, which is that AI is still going to be really
13:13important, that more people are going to be using it and also using it more.
13:18If that holds true, then the bubble matters less.
13:22I think the reason a lot of people think we're in a bubble is just because of the circular nature
13:27of these deals.
13:28And so much spending on AI infrastructure.
13:31Do you think that we are underestimating the amount of compute that we need?
13:37Or is this a worthwhile investment that will turn out?
13:44Yeah, I think both things can be true.
13:46So from what I see is the demand for AI infrastructure is incredibly high.
13:51There's always a lot of demand.
13:54That doesn't mean that some of the projects that exist should exist or should exist at the scales that they
14:01exist in.
14:02I think the question that's happening is really less about, is it worthwhile to deploy a lot of GPUs and
14:11have this capacity that's available to people?
14:14I think most people would agree, yes.
14:15But there's healthy conversations about, is it funded the right way?
14:20Is the financing of this investment in infrastructure, is that being done the right way?
14:26And that's a great question.
14:28And for me, like going back to, you know, you asked me why SmartBird.
14:31This is what I liked about the SmartBird opportunity.
14:34It was a way to build infrastructure without those pressures of getting really big, really fast, with billions of dollars
14:42attached to it.
14:43But actually building it more incrementally with that discipline of how much infrastructure do we want to deploy.
14:49It may seem really small to a lot of people, right?
14:52But it's still meaningful to the customer.
14:54What part of the market feels most frothy to you right now?
15:03There's a lot of companies that are doing something AI-related, right, where it's not very clear what it is
15:15that they're doing that's differentiated
15:17and that is a meaningful value proposition for a customer over a long period of time.
15:23That's always a bad place to be.
15:26So those would be the ones that would be careful about.
15:29Yeah.
15:29Well, you've been very nice to put up with all these stock tarts.
15:34But I want to end with a lightning round to get your thoughts on a couple things.
15:39Sure.
15:40So what is one word you would use to describe the last couple months?
15:46Exciting.
15:46Okay.
15:48In one sentence, what is the most common misconception about AI infrastructure?
15:56That it's all about GPUs.
15:58We actually do a lot more than GPUs and even a lot more than compute.
16:01It's a lot of, you know, networking, storage.
16:03It's really about having a holistic system that a customer can use to do AI.
16:09What is the first role you hired for when starting an AI infrastructure company from scratch?
16:20Somebody who really understands the technical build out.
16:24Everything that we need to put inside the data center to make a customer happy.
16:28Who is another CEO that you most admire?
16:32Ooh.
16:33I have to say Jensen.
16:35Okay.
16:35Jensen.
16:36Yeah.
16:37And what's another company that's gone through a very radical pivot that you most admire?
16:45Oh, that's an interesting one.
16:52In some ways, Google.
16:53I think they've reinvented themselves, you know, a little bit more incrementally, but
16:59in ways that were very meaningful over the years.
17:02And do you think they're doing that again now?
17:04Yeah.
17:04Absolutely.
17:05Absolutely.
17:06I think every company needs to reinvent itself every now and then.
17:10They don't need to do it as drastically as the old birds to the smart bird transition.
17:16That's probably not the preferred mode for most of them.
17:20But I think any company that does not pay attention to the new technologies and what that new technology
17:27is going to mean to society is probably missing something.
17:32Yeah.
17:33And to your point, Google's stock is down double digits today.
17:35You've actually outperformed them.
17:37So there you go.
17:38There you go.
17:39There you go.
17:40And you still haven't worn all birds.
17:43What is your favorite kind of shoe?
17:46Probably something similar to what I'm wearing now.
17:49Yeah.
17:49Which is what?
17:50I don't know.
17:51Should I make their stock go off?
17:53Sure.
17:55If they become an AI company.
17:57Yeah.
17:58Yeah.
17:58Exactly.
17:59What is your most contrarian view about AI today?
18:04One that many people might disagree with?
18:10I think a lot of people only see the tip of the iceberg in terms of what AI can do.
18:18And so I get annoyed when, you know, I tell somebody that I do AI and they ask me something
18:25about chat GPT.
18:27It's like, okay, but that's like one little thing that AI can do.
18:33It's a great thing, but that's not everything about AI.
18:36I think most people, unfortunately, are not as exposed to how AI can be used in real world
18:44applications that are actually useful and meaningful.
18:48And when I was the CEO of DCEI, my last company, that's all we did.
18:53Like we were really heavy into more mission oriented AI and AI that was being used by scientists
19:00and people doing research.
19:01And I think that was really good.
19:03I wish more people could see that.
19:05And I feel like everyone now is very much returned to office.
19:08We just heard that from Steven.
19:09You guys are fully remote.
19:11Why do you think that is the best way to go?
19:14A lot of the work that we can do is either virtual or it's in the data centers actually doing
19:20the build.
19:21I think this is, as an AI infrastructure company, we spend a lot of time actually building things,
19:26which I actually love.
19:28So if my people are sitting in an office, something's wrong.
19:33That means that they're not actually building systems for customers.
19:37So that's a big part of it is basically meetings we can take remotely.
19:41If we're working on a deployment, then they're usually with the customer at the customer side
19:46or the data center side.
19:48Awesome.
19:48Well, thank you so much.
19:49I'll see you guys in a little bit for our reception with Trak and the CEO of Hebbia.
19:54That's coming up in about an hour.
19:57But thank you so much for joining me.
19:58Thank you, Ben.
19:58Thank you a lot.