Atlassian CEO Mike Cannon-Brookes：为什么一切都被高估了，以及我们是否处于 AI 泡沫之中

Reasonable men live within the boundaries, and so you need unreasonable men to change things and move the world forward. One of the only predictions I think I would make accurately is that 10 years from now, we'll look back and say, “Man, you remember that era? That was crazy.” Most things are vastly overvalued.

Five years from now, we'll have more engineers working for our company than we do today. We'll have more software developers working for our companies. We will create far more. They will be more efficient. My favorite quote from Tobes is that he once told me, “The founder's job is to fight the entropy of ambition.”

Mike, I'm so excited for this, dude. I'm one of the biggest Atlassian fans, so thank you so much for joining me today, dude.

1. Lessons from 20 years as co-CEOs with Scott Farquhar

Thanks for having me on, man. Pleasure.

Scott calls you—and this was from one of your team members that I heard—the unreasonable man.

He does.

Why does he call you the unreasonable man, as a starting point?

I know it's after, I think, George Bernard Shaw's famous quote that “all progress depends on the unreasonable man,” because the unreasonable man is the one who doesn't believe in boundaries, makes changes, et cetera—sometimes for the better, sometimes for the worse. Reasonable men live within the boundaries, and so you need unreasonable men to change things and move the world forward. It's always been his nickname for probably 2 decades now.

Do you think you're an unreasonable man?

I think it's not an unfair term. It sounds very negative, doesn't it? In the way that he means it, yes, it's probably pretty apt. I tend to get frustrated when things don't work the way I think they should, and then just try to change them because it seems like the logical thing to do rather than accepting that they can't work, right? Sometimes that leads you into trouble, and other times it leads you into doing things that hadn't been done before.

I'm a massive fan of David Goggins, the Navy SEAL, who basically tells you to continue through suffering. He says that the worst thing is to become civilized. A civilized person is mediocre and lethargic, which is very strange for a European to be saying, but I totally agree with you in terms of the benefits of unreasonable men.

You said that you get frustrated when things don't work the way you think they should. What's most pressing to you that doesn't work the way you think it should today?

Look, I think the people who come with a systems mind and create technology, or get into technology because they see its creative potential, are inevitably thinking, “I can take this thing and that thing and put them together and make that. Why hasn't anybody done that yet?” And then they just go and make that thing.

Sometimes it works wonderfully. Sometimes it's horrible and nobody knows about it. So, I think there are plenty of things that don't work in the world. That doesn't necessarily mean they don't work; they could be better, right? Atlassian is in the business of building technology, and we're trying to make better technology every single day. There are plenty of things that we haven't done yet that we want to go do.

We're obviously in the AI era. Lots is changing there. Lots is changing in the global energy world and climate world, and I still don't understand why my dishwasher is so complicated. So, we can go any which way.

Dude, I've got no idea how to turn on the dishwasher, so you're still ahead of me. It's fine.

You and Scott were 2 of the most famous co-CEOs and really pioneered the structure in tech for a long time. Obviously, you're now a solo CEO. I'd love to hear your reflections on your biggest lessons from being such successful co-CEOs for years, and what worked and what didn't with the benefit of hindsight.

2. Why design, not data, will define the AI era

Look, there's no doubt Atlassian wouldn't be where it is without Scott—probably without both of us, without the relationship. It's really hard starting a company. It's really hard building something. It's probably even harder doing it on the far side of the world from most of the other people who are doing similar things. Having 2 equal parties, I think, in motivation and experience has always been a really important part of what made it successful.

By equal, yes, equal shareholding-wise and equity-wise, but also equal in terms of age and life stage, equal in terms of naivety, experience, and wisdom. I think that really helped, because you'd go into a room and be like, “I don't know how to do this,” and he'd go, “I don't know how to do it. Well, we've got to figure it out.” You have someone who's at a very balanced stage of all of those sorts of things.

We both started having kids around the same age, and we both got married around the same time. That really helps because when you have a kid, you're like, “Oh my God, this is really hard. I didn't get any sleep last night.” And that goes for a long time. It's really helpful to have someone else who's in a similar situation, and you can talk through it and work through it. We've juggled and balanced for decades.

I think that's one thing that really made it work: the equality, the balance, and that sort of shared world. I think the second is that he's far better at this than I am, and I know he said the same thing. I think it really helps to think the other person is just far better than you are at pretty much everything, because you're both stretching to try to be as good as the other person, if that makes any sense.

I know it's been said that, in life partners, you want to believe your partner's out of your league, because then you're like, “Oh, I better show up.” If both partners believe the other person is out of their league, it really makes sense.

David Goggins is a good analogy for Scott in terms of his ability to tolerate shit, get through it, and keep going. The man works so hard. He sets a high bar. So, for 20-plus years, I felt I had to keep that high bar, right?

Lastly, look, we're good friends. You've got to laugh a lot about these sorts of things. You're going to go through all the ups and downs over many, many decades. I think we're both pretty grounded people and just had a lot of fun with it. We've been on the ride of a lifetime, and you just have to enjoy that, right? It really helps to be able to have a beer, hang out, and enjoy it.

Two of my dear friends, Alex Nordstrom and Gustaf Sodestrom, have just become the 2 co-CEOs of Spotify in a really interesting new structure, obviously replacing Daniel Ek. What advice would you give to them entering this structure, knowing all that you know?

3. Becoming solo CEO & making bold AI-era moves

Oh, knowing a bunch of those people, I'm going to get myself in trouble.

Don't worry. It's just between us, dude.

I think the advice would be that you've got to be as open as possible, dealing with conflicts and dealing with perceived conflicts. You've got to be very careful to try to keep balance, right? You've got to have balance of the force, right? There's a yin-yang kind of element to any of these partnerships that has to work.

At the same time, you have to have a swim lane. That's where the magic is, somewhere in both of those things, right? I once described it as Scott and I probably being somewhere between 60% and 80% overlapped. If we're 100% overlapped—like we do exactly the same thing—well, there's not much point in having 2 of us. If we're 0% overlapped, you're going to have a lot of conflict, right?

You need to have enough shared space in the middle where you understand what the goals are, but also a time when you're like, “Well, I'm doing my thing and you're doing your thing,” right?

In Atlassian's history, at various times, both of us were in the whole business, where someone was off on sabbatical or getting married or something like this, for periods of up to maybe 6 months. At other times, we moved the various functions around. I think both of us probably ran all the functions for various times, anywhere from a few years to—I ran sales and marketing for 13 years, which was probably the longest of any single function staying with a single person. Then I moved it to Scott for 4 years.

There were swim lanes. You have to have an understanding of roles and responsibilities to make it work, but you also have to have quite a lot of shared context and shared understanding. That equality was really important.

When did you most vehemently disagree, and how did it resolve itself?

Well, look, firstly, our first shareholder agreement actually had in it that if we couldn't agree on something, we'd go to a game of scissors-paper-rock, best of 3. We never got into any bigger disagreements when that shareholder agreement was actually governing the company. Don't worry—as a large public company, that's no longer in the bylaws—but it was for well over a decade.

I always knew I'd lose this scissors-paper-rock game, best of 3, so therefore we never got into any disagreements that required the use of that clause. I think we've disagreed over acquisitions, strategic directions, et cetera. I've always lived by the maxim: if I can't convince him that something's a good idea, it's probably not. That is a really good marker that something is probably not a good idea, and vice versa.

I think he would say probably the same thing, which meant, if you did disagree, maybe you tried to lobby the other person and convince them for a while.

Well, if you couldn't convince him it was a good idea, we didn't do the thing. It worked pretty well. There were no real major disagreements that were epic, singular points where the business went one way or the other. We've been in the same boat for a long time.

I remember him telling me the story of actually being on his honeymoon, being on safari, and being completely off-grid, and you having to get hold of him. I can't remember what for, and Kim had to go to some random-ass village. I was like, “This is why I don't leave London. I get hives when I leave a delivery zone, so it's not a good thing for me.”

Entirely true. We had our first major security incident decades ago now, and I needed to contact him for various legal reasons. He had to be notified, so he landed in a bush plane somewhere in Africa and got a radio message, I believe, saying, “When you can, call Mike.” He was hundreds of kilometers from anywhere.

We sorted it out. It's a good example of where you need co-founders. It's better to have 2 of us in that situation. We had a replacement.

Going back to the unreasonable man, and you now being the sole CEO, when I reflect on your last 18 months, respectfully, and I hope this isn't rude of me, it's probably the most aggressive but also obvious example of founder mode in action. Do you think that's a fair description? Was it a deliberate, “I want to make some big, bold moves to show the market I'm here to really play”?

I'm going to violate the old media rule: never disagree with journalists. But you said you're not a journalist, so that's all right.

And if you disagree with me, please do. I'm not a journalist. Yeah.

I don't think so. It's interesting. I don't think I've seen it characterized that way. There's certainly no intention to be like, “Let's go make some bold moves.” If anything, I ring him all the time and say, “Hey, what do you think about this idea?”

I get all the credit and all the debit now, so that's interesting. Anyone who knows about technology companies knows it's a little bit like when a new prime minister comes in in England or Australia, right? Suddenly, interest rates go up or down a month after they've come in, and it's their fault. They're like, “Oh, yeah, I wasn't here for the last 5 years of running the economy, so suddenly I get the credit.”

Many of the things that we've done over the last 18 months were planned many months, years, or quarters in advance of that.

But if I push you there, and again, you can disagree with me. I'm not a journalist. I'm a VC, so you can legitimately tell me I don't know what I'm talking about, and then you would be right. But the sports investment in Williams—I love Williams, by the way. We just did a show with James. He's fantastic. Awesome dude. The acquisitions, a browser company, DX, some of the smaller ones—these are pretty big, bold strategic decisions that I imagine were not longer-term planning.

Yes and no. Some of them are certainly longer-term planning. Other ones, I would say, again, to the characterization, it's not a desire to stamp some sort of “Now Scott's out, I can do the things I've never been able to do beforehand.” There's literally zero of that, right?

There's no doubt the AI era, whatever you want to call it, in the last couple of years has been a little bit busy, and there's a lot going on in technology. I think for every company, it's probably a more active period. If you're really trying to compete, if you're really trying to create, this is a time when, A, you should love it, and B, it's a little bit of chaos, right? It's not a linear period. It's a period where you need to invest in things and make some bets.

It's a very creative period. I tell people all the time, this is the reason we got into technology. This is amazing. This is one of the most exciting periods. We will all look back at this. I guarantee you, one of the only predictions I think I would make accurately is that 10 years from now, we'll look back and say, “Man, you remember that era? That was crazy. That 3-, 4-, 5-year period was just a complete up-and-down turmoil.” There'll be some things where people will say, “Remember that thing? We all thought that was going to be huge, and it was nothing,” and everyone else will be like, “Yeah, no one saw this thing coming, and it was massive.”

I do think it is a very creative, Cambrian explosion period that will create change. I do think it's a time when, for founder-driven businesses, you need to make some decisions and take some real actions. That is certainly true.

I do think some of those decisions have been in planning for a while. In terms of acquisitions, we acquired Loom 3 years ago now. It was a massive acquisition. It was a bold bet. I don't think it's particularly different from the 2 we've just done. It's done very well. It's a fantastic product and a fantastic team. They're cranking away, and it's been a fantastic thing for the AI era. As we said at the time of the acquisition, it was one of our more prescient choices.

4. Are tech valuations insane? Mike’s investor reality check

You said there about the craziness of this time. It is, and I find that quite hard because I'm looking at this as an investor, going, “What is sustainable? What is not sustainable?” I'm trying to assess durability in a world of complete unknowns. How would you advise me, Mike?

I think you've got some very difficult choices as an investor at the current time. Most of the things are vastly overvalued. Probably true. Some of the things will be worth far more and are undervalued, and the far more is probably outweighed by all of the things that are overvalued, as classically happens.

For all the dot-coms that went bust, Amazon turned out to be quite a good business. That makes it particularly hard in this era to be an investor. Secondly, we aren't in an era where we have a durable business model for a lot of these things.

The old argument that some AI company gives $100 million to a model company, who then gives $150 million to a cloud provider, who then gives $200 million to NVIDIA to buy chips, and everyone's like, “Look how much revenue we've all got,” and you're like, “Yeah, but you're all losing $50 million on the way through.” It's a pretty good analogy, but maybe they'll make it up in scale, right? I don't think anybody really knows.

We know that it's a foundational technology. We know that it changes so many things. Where the monetary value and the fundamental value lie at the far end of it, in terms of how the business models shake up, I don't think that's known. If you think about the search era and advertising and all these things, right? Anyone who tells you that they know it is probably not telling you the truth.

It does not mean that there isn't real value there, right? It does mean it makes it very hard for you to be an investor. I'll agree with that.

How does that then reflect on how you behave? Because you have to play the game on the field. You have to make bold bets. How does that reflect on what you do?

I think what we've tried to do is a couple of things. Firstly, we've said for 2 and a half years now that we need to make bets. We need to make choices or form opinions about how we think things are going to evolve, but we need to be willing to change those.

We have various hunches that we've made where we've said, “Hey, we think over the next 3 years this is going to be what happens.” But about every quarter, we reassess those sets of fundamental choices. I can go through a bunch of them, but I think that's one thing that you have to do as a business. You have to be willing to walk away from those hunches that you have, but at the same time, you have to make them. You can't be in this era of, “Oh, I'm not really sure what's going on in all these different areas.”

Secondly, I think you have to listen to your customers very deeply because in an era where everything's swimming around and there's a lot of noise in the world about what is and isn't, oftentimes you've got to find the stable thing, which is to go back to the customers and say, “Do you feel this way?” Talk to enough of them that you get internal conviction about what is or isn't happening.

You've got to talk to the right customers, and you've got to talk to enough of them. That is definitely true. Those are the people that, fundamentally, for whatever business you're in, will hopefully tell you and help be a rock in a storm of change, let's just say.

I think you have to make some bets, right? In the way that you used it, you have to take some risks and try to change your position.

You said about having a thesis or a belief on a market or an evolution and then revisiting it. What belief did you have that proved to be wrong, and how so? What belief did you have that proved to be right? I'd just love to understand 1 on each side of the table.

We made a couple that I think are proving to be right. I don't know if anyone in AI could say that they've been proven to be right. That's probably a little bit arrogant at the current time.

Firstly, very early on, we thought there would likely be multiple foundational models competing. Hence, we've built our technology to be multi-model on purpose in almost every way, and we've decided that we were not going to make foundational models. It's unlikely we'll be able to compete to train our own models. However, we have to get good at something, and what we are good at is adopting and delivering the value of those models in software to our customers very, very quickly.

So, if you think, “Okay, there’s going to be a new model about every 3 months, and there are going to be 4 to 6 companies competing,” what we need to be able to do is pick up a new model, test it, work out whether it’s better in a certain case, deploy it, and get it out to customers. That’s an expertise that we need to build, and we have built it. That’s proven to be, I think, a really good bet.

A second one for us is that it’s really all about design. Yes, data is really important. Yes, models and fundamentals and all these different things are important. I think what is underwritten is the importance of design.

If you look at any lowercase d—and I don’t mean graphic design, like how it looks or the colors—when you have any major technology or cultural transition, we’re back to an era of fundamental design. We’re rethinking entire swaths of the world.

The trite example, actually, is pull-to-refresh on your phone, right? We never really had this design device until we had phones, and then suddenly people said, “You know what? One app did it, and everyone copied it because it worked really well.” In AI, we’re facing that very much more, right?

I don’t believe the world will just end up as a chat box. Otherwise, we’d all just use the terminal on our computers. But it is an entry point, and so there’s a lot of foundational design being done—really fascinating, good design all over the place. Even where some of it fails, like most good design, you’re like, “I see what they were trying to do.”

We have very heavily invested in design in the AI era. In an era where, arguably, software gets cheaper to create—we can argue about that, maybe—but there will certainly be far more software, which is probably a fairly safe assumption, and far more technology. What differentiates it is probably how it feels and how it works. That’s very hard to copy, right?

I think that’s another one of our big bets that we’ve made internally. We’ve really beefed up the size, the quality, and the talent on a design team that was already pretty world-class to start with. We’re doing a lot of foundational design around delivering AI to end users and customers that don’t have to understand what the words “probabilistic” and “deterministic” mean, or what a model is, and all these other things. They just want to do something.

When I hear you talk about the centrality of design there, I think about Satya Nadella’s statement: “Business applications will collapse right in the agent era. They are essentially CRUD databases with a bunch of business logic.” How do you feel about that, reflecting on your centrality of design?

It’s very easy to make big statements in the current era because you can’t be proven wrong. Hence, those statements could theoretically have some value of being right.

Do I believe that all business applications—maybe all applications, full stop? I guess you could argue every piece of an application is just a CRUD piece of software. Email’s just the same thing, right? I don’t know if it’s a business application or a consumer application. I still believe there’s a role for software to solve certain purposes.

There’s an old adage in enterprise software that every single piece of enterprise software can be replicated with email and Excel. It’s quite true, right? You can do any application you want with email and Excel. You can replicate any communication app. You can replicate any business-process app.

There’s a reason we don’t all just sit there with email and Excel. They’re still very used, very popular, right? But at the same time, they are not used for large-scale things. We have HR systems and CRM systems and project-management systems, and lots of communication systems from Zoom and Slack and all the things, right?

They’re all very specific versions, you might argue, of email and Excel, but they serve a very specific purpose. What we’ve seen over the last 10 years is the opposite. We’ve had a Cambrian explosion of SaaS apps to do more and more and more specific tasks, but they do the task much better.

I don’t think we know, in the AI era, if all those apps just disappear into some godlike Siri agent that we just talk to. I’m not a believer in that worldview. I don’t think it’s likely to happen.

I do think AI is going to have to change all of those applications in massively fundamental ways. That may mean some of them end up disappearing and others end up profiting. I can believe in that worldview. I don’t think that they’re all just simple CRUD apps. I think that’s usually someone talking their book.

I very much buy into, I think, a statement likely attributed to Eleanor Roosevelt: bad discussions are about other people, mediocre ones are about events, and smart discussions and good discussions are about the future and ideas.

When we look forward, you said that you don’t believe the chatbot is the interface of AI in the future. What do you think is the interface for AI in the future?

It’s a good question. Look, I think that’s what we’re all trying to create, right? It’s one of the fascinating questions around today. It’s the reason why fundamental design is such an important skill and talent, I would argue, in the next few years—maybe forever—but right now, it’s hugely important.

I think that interface is going to be, probably, far more evolutionary. So, I don’t know if there will be an interface. It might be that AI is rewriting interfaces. In theory, AI can generate some code, write the code to create a new interface for every single screen you look at, right?

I don’t think we’ll probably see that. It’ll be too discombobulating for users. They’ll get lost. They’ll be like, “Wait, everything’s moving all the time.” If you watch users in any user study, they like knowing that, “I go to this button, I click the thing, and that happens, and I go to this thing, and that happens.”

So, they’re going to want stability in their interfaces. They’re not going to want this dynamically changing, always-changing interface, but we can generate new interfaces pretty quickly. I suspect it might look a lot more like today than we think, but it will be a lot smarter. It will do very different things, and there will be a lot less click-to-value ratio.

There is no doubt that a lot of software requires you to click or type to indicate what you want to do. A lot of them take the other extreme: prompting and writing into a text box. At some point, my prompt gets so long to be specific about the thing I want, I’m like, “Just give me a damn button that I can click and a slider to move and some UI interfaces.”

Somewhere between the two is probably where it is, right? It may be a combination of modern existing software UI and prompt-based layers. We’ve seen lots of people playing with this sort of technology, where you can customize software with a prompt.

Microsoft Word has a lot of buttons at the top of it. And you say, “You know what? I want to be able to edit a prompt in Microsoft Word that says what type of person I am,” and that changes the UI.

I don’t understand some of the toolbars there. I’m just completely lost as to what they do. I never use them, but I’m not a lawyer. I don’t live in that world. I’m like, “Well, just turn all that stuff off, because I’m going to tell you who I am, and then it’s going to redesign the interface for me.” I can see that totally happening.

But again, this is back to why we need fundamental design in so many of our day-to-day applications, and what makes it such an exciting era to be building technology.

Does the democratization or the cheapening of software creation make software more or less valuable? Does it make design more or less valuable? How does it change discovery with an infinite supply?

I do think it makes design more valuable. Good design is a scarce resource, which probably gets more valuable, and I think good design is quite scarce.

5. The future of engineers in an AI world

There’s no doubt in my mind that we will create far more technology, right? We often get asked, “Do you think all software developers are going away?” I said no. I was speaking right before Matt from AWS, and we both violently agreed that 5 years from now, we’ll have more engineers working for our company than we do today—more software developers working for our companies.

We will create far more. They will be more efficient, but software—technology creation—is not output-bound. It’s like the demand-supply dynamic is very off the charts.

What I mean by that is we don’t have a shortage of ideas for new technology that we want. Find me a software company or a technology company that gets to the end of its roadmap. The roadmap is unlimited in length, and we will keep coming up with new ideas.

Human creativity means we will come up with new ideas for things that we want to build into our software. Maybe crap ideas, maybe good ideas. We’ll have to build them and test them and see.

I like to be an optimist and think we will end up with far more technology, firstly, and secondly, far better technology. If I can afford, if things are “cheaper to build,” or more efficient, maybe I can build them 2 or 3 times to get it right.

Plenty of technology is built once, and you use it. It doesn’t really work very well, but it works well enough. Some team somewhere was like, “Man, I’ve got to move on. I’m out of time. I’m out of budget. I have to move on to building something else.”

Maybe if they’d had 3 attempts at it, they would have made it better, right? So, we’ll all live, I hope, with a world of better technology and far more technology. There are a lot of problems to go.

Well, I do want to jump on a lot of that, but we see Cloudflare, Salesforce adopting—Microsoft even adopting—vibe-coding environments that replicate Lovable or Replit. Have you internally discussed doing the same with Atlassian?

Do you mean using those vibe-coding environments in—

No, in terms of creating an environment where, within an Atlassian Home, people can vibe-code within large enterprises.

Sure. I'm in Barcelona, and we're launching one at the moment. So, yes, I don't think this is going live in the next 24 hours, so that's fine. Yes, there's no doubt that that's there, but again, it's about focusing.

What's the thought process behind that? Because you've got to then believe that vibe coding as a category will be a sustainable and enduring category in software creation. Do you believe that?

It might depend on our definition of vibe coding. I believe that there is a large group of people who we may or may not call software developers who will want to be able to create technology of a certain quality or level, or that meets a certain set of standards—maybe their design standards, software security standards, whatever it is. I'm not sure whether we'll call these people software engineers. I don't think it really matters. I think it becomes one of these annoyingly not useful debates.

If you ask me, at Atlassian, I will end up with far more generalists. I will end up with a wider organization, and people will do far more things. Their jobs will hopefully be more fulfilling because of it, right? I will have people in finance who are quote-unquote vibe coding an application to solve a financial problem for themselves, where previously they might have tried with Excel, filed a ticket with IT, or just wouldn't have bothered, right?

They're writing Python code to do crazy, complex analysis. The cost of that code is so small that they might do it three or four times, maybe throw it away because it didn't work, right? That is not what I would argue is software development. That is a financial professional who's trying to get their job done and has a new set of tools to go get it done. Marketers making websites, you might argue, is kind of the same thing.

There's lots of these examples where it's people whose job description wouldn't say, “I'm a developer,” but who are now making technology. That's a great thing. I don't think it means there's less of a role for the core technologists to play. In fact, they're probably building more and more platforms.

We are trying to build, for our maker audience—we have a huge community of app vendors, creators, and customers who make technology apps that run on top of the Atlassian platform. We want to help them make high-quality applications that solve business problems on top of the Atlassian platform at the lowest cost possible. Now, you can do that by writing code for a very complex application that does a lot of things, or you might vibe code something really simple.

If by vibe coding you mean largely writing prompts and spitting out an application, and if it fits into the design language of Atlassian, runs well, is secure, and is compliant, that's great. It'll probably mean our customers make far more apps per customer than they ever did beforehand. A lot of them may be throwing them away, right? That's okay. That's fine. That's great. There's going to be a big spectrum.

The spectrum is a barbell, in some respects, between what we wouldn't traditionally consider developers and the really valuable engineers and developers. The question becomes: You said we will have more software engineers. Will we have the entry level? Will we have the kids studying computer science who come in and are decent, but definitely aren't 10x engineers and aren't specialists? What does the future look like for them?

I think we'll have far more of them. Maybe that's a bold bet. We're hiring more grads this year than we did last year, and more than we did the year before.

If you think about it, it's purely an economic question, which is dangerous when it comes to talent. If someone was going to be a 10x engineer 10 years ago, I would argue that person is probably still a 10x engineer today. Maybe they're a 100x engineer because they've got all these new tools, right? That's probably down to the person, their learning style, their ability to ingest information, and how they solve problems, et cetera.

Maybe the average engineer is going to be a 10x engineer, and those 10x engineers are now 100x engineers. It will come down to how they solve those problems, firstly. Secondly, I still have a bandwidth constraint on the number of people I have in R&D and engineering to build the ideas that the business could profitably build. That is still a constraint that I believe will be forever, but it's certainly a constraint that we've lived with for a long time. It's not going away today.

Lastly, we've seen this with Loom, fascinatingly, in a very different context. There's a good chance that those graduates come in with a different view on what it means to be a software developer and shake up the existing world of talent in a positive way for my business, where the existing talent either figures out, “Man, these grads are doing things better than me. I better copy that,” or maybe, “I should go do something else,” right?

The reason I say we saw this with Loom is that one of the ways Loom gets into businesses—Loom being a video and communications tool—is through graduates. Young people coming into businesses use a lot of business tools and are like, “Man, this is so inefficient.” Young people, the TikTok and Snapchat generation—I can't count myself, unfortunately, in that generation, I don't think anymore. I've never really understood Snapchat, but I try.

They are natively used to FaceTime, video calls, and voice notes. This is a very common communication format for them that, maybe, you might argue someone who's been in the workforce for 30 or 40 years is not used to. When they come into a business and use Loom to communicate, the business goes, “Man, those people are more productive.” The entire business starts communicating that way, or a majority of it becomes a norm and a standard, right? Because they come in with a different expectation.

I do wonder if, as a grad today—say you're halfway through your university degree and you're about to finish computer science—you're surely using all of the AI coding assistance and everything else. You are hyperproductive compared to what a grad would have been 10 years ago. You're going to go into a business, and that business is going to say, “Well, that talent is amazing. I need to do more like that.” So, they're going to bring in those skills to businesses.

Mike, can I ask you: When you look at the engineering team today at Atlassian, what are the dominant tools that have changed productivity meaningfully?

I have to talk my own book, but Rovo Search. Look, we use our own tool, Revo Dev. We have many code-creation tools, so we use Revo Dev, Cursor, and Copilot. We use a lot of things. We've been very liberal. They're quite cheap, these tools.

Would you pay 10x the price of those tools?

No, probably not. I mean, maybe if they gave me 10 times the productivity.

I've asked 10 CEOs this at least, and you're the only one who said no.

Because I think the competitive vector between them is not going to let you do that.

In terms of the switching cost—

If one of them charged 10x, you would just switch to the other one. Secondly, they're all good at different things. So, there's a reason our engineers have access to all of them, and they choose to pick up a tool for certain reasons. Sometimes it's preference. Sometimes it's for the task at hand or the type of thing they're trying to do.

Most of them are pretty good at greenfield development: I'm blank and I want to start something. Some of them are very bad at long-form work: I want to manipulate this piece of code across 500 repositories around my organization. Some of them are very bad at ingesting a large codebase. They will all get better at different things, but there will be, like anything, preferences and styles and strengths and weaknesses.

Secondly, coding is a pretty small part of a software developer's job. Different surveys say 10%, 20%, or 30% of the week is spent writing code. You might argue that in the best companies, it's close to 50% of the week spent writing code. There are meetings and finding things and finding people and finding APIs and understanding the codebase.

Search is probably about the same amount of time as writing code for most software developers as a professional. Debugging, quote-unquote, at the large scale: something's gone wrong in production; you're getting an alert, an error. Okay, let's figure out what's going on.

Yes, there will be AI tools for that as well. There are plenty of AIOps tools. We sell one. There are many others that are trying to help you with the operation of software, the running of systems and services. I think that will get even more complicated because, if you have far more technology, you're going to have more services.

Instead of running, say, 100 different services as a medium-sized startup, you're going to have 1,000 services. So, the permutations and combinations and finding problems and gremlins are going to get that much more complicated. The answer to all these questions can't just be, “Throw more AI at it.”

You end up in this weird Gordian loop of, “AI is going to solve the problems AI created with AI,” and it continues down the AI path.

Well, that's what venture investors are telling our LPs at the moment. So, just don't ruin the money train for us, if that's okay. We've never had a technology disruption or generation before that wasn't going to solve every problem in the world.

You mentioned the switching cost. There's a great book by Hamilton Helmer, Seven Powers. If you haven't read it, please let me send it to you as a gift. It's the best business book you'll ever read. Bluntly, one of them is that it's incredibly difficult to switch from a service, as a core trait of a great business.

So it could be Salesforce. It's a fucking nightmare to switch from Salesforce. That's why no one does, despite it being shit.

Yeah, Mark Benioff's a friend. So, if you're listening, Mark, I'm so sorry. Don't kill me, but it's true.

He's built a great business then, right?

He's built a great business.

Yeah. But if this is true, that switching costs are actually incredibly low, are these fundamentally good businesses?

Look, I think a lot of things in AI have been built very fast, both in terms of the creation of the technology behind the business and the customer base and everything else, which almost by virtue of that means the switching costs are quite low at the moment. You cannot build a business fast that has high switching costs. It doesn't logically make sense. Do they have high switching costs in the future? That's a very different question from whether there are high switching costs today.

One of the fascinating things for me about AI is that we navigate this disruption, this period of creativity, while large language models have been dropped on the whole world at the same time. It's like someone just went, “You can all have these for free. Go,” which is something I don't believe we've seen in technology before. It's very unique.

It took a long time for mobile phones to permeate the world. It took a long time for PCs to permeate the world. It took a long time for these things. It's very different to have 1 billion technically literate people get access through cloud providers to it. It doesn't matter: OpenAI, Anthropic, Gemini, Llama, Grok—everybody. That's why you're seeing such vibrant creativity, which is fascinating.

It also means that the switching costs of almost everything being built right now are quite small, and they will have to build up, one would argue, to build sustainable businesses. How those switching costs build up, though, is much harder to see. We don't really understand that, because if everyone can copy everyone really quickly, you're going to keep those low.

The switching costs are going to end up being value delivered to customers, data to some extent, workflows, familiarity, maybe interface similarity. Again, why do you use a certain app on your phone? I'm just used to it. Do I go look for a better one? No, it's good enough, and so I don't. I don't think we really know where the switching costs are yet in some of these particular AI-driven businesses. It's the old sort of, “Does HBO become Netflix before Netflix becomes HBO?” kind of loop.

You've got a lot of these existing businesses like us. I think we're arguably an existing business, 20-plus years in existence, that has a high R&D investment and the same access to all the models. It's about how much we can infuse AI into our core platforms and experiences for the 300,000 customers and tens of millions of end users we have, faster than these new companies can come along.

It's a classical technology race. The best statement ever is, “Can the startup acquire distribution before the incumbent acquires innovation?” I think that's this continuous cat-and-mouse game that I often think of.

People often throw out, “There's no switching cost between X or Y.” Once you get past that and go, “Okay, well, then the fundamental question is, how do you build defensibility post that phase?” The next criticism is, “Well, Mike, come on. They've all got shit margins. They're all negative-margin businesses. They're crap. They can't be good businesses.” How do you respond to the critique of the low-margin nature of relatively all of the AI tools that we're using?

We're too early to really understand what those margin profiles are. We don't understand the monetization models for a lot of them yet. They keep changing pricing schemes every 3 months, so we don't understand the monetization model or really where the relative value is, or the time degradation of model value.

The whole question of, “Hey, I've trained a model. It cost me $1 billion. Am I going to get that $1 billion back in the 9-month period before the next model's better and it costs me another $1 billion to train?” These have all got to be sorted out. I can't tell you how the world is going to sort them out, but I can tell you that they will be sorted out in some form.

Margins are probably built in these businesses with customer value delivered, right? There is a partnership and a relationship with your customers, and a faith that you will continue to deliver value that, in some ways, builds margin over time in different 7 Powers manners than just pure switching cost, which is often about a somewhat captive kind of notion.

Then you get to, well, where does the fundamental value live? In any technological change, the value can shift around. You might argue, well, the chip vendors are going to capture all the value. Some investors believe this at the moment. Possibly, you might argue the power producers that are upstream from the chip vendors are the people doing really, really well.

If you're selling electricity right now, it's a good business to be in. I don't suspect it all ends up flowing all the way up to the power companies making all the money in the world. That doesn't seem to make any sense to me, because there will be, again, design, aesthetic and style all the way down at the other end that build switching costs.

There are lots of applications I use where maybe I pay more than a sheer economic value because I prefer the design of that application. I prefer the way it makes me feel. I prefer the way it makes me productive. It's not all down to spreadsheet value, like some people would have you say.

If you're just an OpenAI wrapper as a startup, I do think you're in trouble. But that might not be the case—we might look back and think that was a good way to start some of these businesses, but they end up being very different. Plenty of businesses made money on mobile without being iOS or Android.

In terms of making money, I think the core question—and not picking on Salesforce, but picking on Salesforce—is the pricing model. We've got so used to per-seat pricing. It's what predicates Salesforce's growth now, bluntly: just an expansion in seats. Does the per-seat pricing model die in the next wave of SaaS, and how do you think about navigating that successfully?

I think the only logical replacement out there—well, there are 2 logical replacements. One is per-value delivered, some unit of value. The other one is some sort of consumption-based model.

Back to my prior point, go talk to customers about how much they love consumption-based models. Generally, they don't love them very much because they're not quite sure how much they're going to consume. They don't want to be in control of their employees. So, the world does trend to a non-consumption-based model in many areas, not in all areas. AWS is obviously consumption-based.

At some point, those consumption-based models tend to flip from something that's clearly transactional, like Stripe or something where you can clearly measure consumption, into another form of payment. We give away unlimited storage; many vendors do. It doesn't mean the storage doesn't cost us money, but it's our job to get enough money from a customer that we can, on average, pay for the storage, and we have the right controls of exception limits and stuff like that.

I do wonder if AI ends up that way around. Value-based pricing, I think, is fascinating on a delivered-outcome basis: “Pay me for every X I deliver.” The problem with that becomes 2-fold, I suspect.

Number 1, you've got to measure X very accurately so that both the seller and the buyer are prepared to agree with the measurement of X, which is a big problem in a lot of these scenarios where people give these pricing schemes. Secondly, if it's, “I can do X cheaper than you could,” that works for 1 loop, 1 lap of the world.

As soon as it's cheap, everyone's going to cut your price. Let's take a classical support argument. The cost per support ticket is $100. If I could get it to $80, you'd give me $10. You'd say yes. In year 1, I would. I'd pay you $80. I'd pay you $90 to get the $80.

In year 2, I'd be like, “But it only costs me $90 now. I already gave you the $10.” Ergo, I can't keep charging you for the discount, because we're going to end up at negative. You're going to pay me for support tickets.

I think we'll end up with some blended model. The answer is probably a nuance and subtlety, like always. I don't believe per-seat or singular models like that are going to go away, but we'll probably have to get used to a little bit more of a consumption-based pricing. I think it's the safest world at the moment.

Imagine I'm Scott. We're sitting down together by a fireside. You've got a whiskey or a brandy or whatever it is, and we're talking about the threats to the business. Again, we mentioned Salesforce. I think the per-seat pricing is a big threat for them.

Southern Comfort, by the way. So, you're drinking a Southern Comfort right now?

6. How Atlassian survives by protecting creativity

I'm drinking a Southern Comfort. We're in a beautiful pad. I'm really trying to visualize this one. We sit down together and go, “What is the thing that we have to protect against?” This is your and my thing that we've built for 20 years. It's a new world. What do we have to protect against together?

I think the thing that we have to protect, from an Atlassian point of view, is our creativity. If I zoom out, one of our earliest maxims about the business is that we wanted to be a multidecade technology company.

Some of the companies we looked up to the most, as nerdy students of technology, were Microsoft, Adobe and Intuit, because they survived multiple decades and now they're all still around. We have fantastic Adobe and Intuit people on our board, maybe not coincidentally, and we still look up to these companies.

They’re still competing today, 20-plus years after we made up this maxim. The reason you want to be a multi-decade technology company, I would argue, is that you have to survive technological disruptions, just like now, right? We started, you’d argue, at the very beginning of the internet, kind of the SaaS era. We survived the transition to SaaS, and we survived the transition to mobile. Whether crypto happened or didn’t happen, I don’t know. Other people can debate that. Now we’re in this AI transition, right? Those are kind of our 3 decades, put super simply.

If you look back to Adobe or Intuit, they had to navigate DOS to client-server, to Windows, to the internet, to mobile, to all these things. What sets them apart, I think, is that those types of companies are always creating. It’s very hard in technology to survive multiple decades with just a really good thing that you can protect and defend. You have to create, which means you have to destroy. You have to move on. You have to change and adapt. I think that’s done pretty well for 23 years. If there’s something we have to protect, it’s doing that.

It’s not going to be winning in AI. That’s going to be the next 5 to 10 years, and I think we can greatly improve our position, right? I think we’re really well set up for the AI era, but we have to go prove that, and we have to do a bunch of stuff to actually show that that’s the case. In the long arc of time, it’s going to be whatever’s next after AI, which seems so far away to think about. I guarantee we’re going to need to create our way through it, right? We’re not going to be able to defend our way through it. We’re going to have to create it.

That probably comes down to people who want to do that creation, who have a place that they want to work, who have a culture that they like, who want to defend the place through creativity, and who have the openness and the environment where they can say, “We need to kill this thing, and we need to go build that thing.”

What would you most like to do, but because of resource constraints, you’re not able to?

I don’t know that I think unconstrained is useful. I’m not trying to challenge the question, either. I think unconstrained thinking is not particularly useful, right? Constraints are what make everything happen. They’re what make games more fun. A football game with no 90 minutes on the clock and no ending would just be boring. We’d just be like, “All right, it’s 405 goals to 403, and we just keep playing forever.”

It’s called Test cricket, my friend.

Five days of the fucking draw. Pat Cummins is still bowling. Fuck. Well, the Ashes was great last time around.

Look, I think if we weren’t constrained, it would be around talent acquisition, right? Finding more and more people. There are lots of constraints when it comes to creative, wonderful people in the world. There are a lot of great places for them to go work, but I don’t feel like we are particularly constrained at the moment as a business.

I can’t complain. I tell people all the time, “Look, we’ve been here for a while. Think about Atlassian as a company with well over 10,000 people, hundreds of thousands of customers, millions of end users, billions of dollars in the bank, and 10,000 people in R&D who are amazing and can build awesome stuff. Now, let’s go build something.” If you started from that point, you’d be like, “Oh my God, we can’t complain about anything in terms of our ability to get stuff done.”

I remember being 3 people, and we were like, “Shit, someone’s got to pick up the phone today. Someone’s got to write some code. Someone has to take out the trash.” Those are really constrained businesses. We don’t have constraints in that way. It doesn’t mean we don’t have trade-offs, but we don’t have constraints in the same way, I think. It’s more useful to think that way: that we can go and build anything we want to build, but we still have to trade off and build the right things.

The way that you just said that was incredibly inspiring—the scale and the size, and then, “Now let’s go build something.” It made me think of a quote once that a brilliant founder told me. I think it was Toby at Shopify who said, “This is a game of who can survive the longest.”

7. Fighting the “entropy of ambition”

You speak with such vigor and energy today, 23 years in. It is a game of who can survive the longest. Has there ever been a time when your energy or enthusiasm waned? How did you get through that, and what would you say to founders who maybe are waning in energy?

My favorite quote from Tobes is that he once told me the founder’s job is to fight the entropy of ambition. I wrote it down, and I was like, “Fuck yes, that is exactly what I have to do,” because the moment we stop being ambitious as a business, all those numbers and things that I just said—we are freaking tiny as a business, right? We are minute, and we have to feel like we are minute. We have to feel like we have to fight for every customer and every user, and we should. On the scale of the global economy, we’re tiny, which means we’ve got a long way to go.

There are certainly times in 23 years where you get tired and exhausted. I think you have to recognize that there’s a physical tiredness and a mental tiredness. Maybe they’re a little bit different. With physical tiredness, yes, you’ve got to take breaks. You’ve got to be careful with the old classical idea that you’re running a marathon, not a sprint. You’ve got to figure out how to be there and deal with little bits of sleep or whatever it is. Some people are wired for doing that, and some people aren’t. I’ve always been pretty wired to do it.

With mental tiredness, I think you have to think about things in eras, and you have to take breaks mentally as well. I don’t know. Maybe I’m not a very good person to answer this question. I have the best job in the world. I’ve always thought I have the best job in the world. If I did something else, I’d probably do largely the same thing.

I think you have to be very self-aware as a founder about what it is you actually enjoy. If you don’t enjoy the journey, quote-unquote, and the ups and downs of the founder journey, it may not be a journey that you want to be on, right? I think there are a lot of people who want to be founders. It’s tough. There’s no sugarcoating it, right? There are days and weeks where it’s bloody hard, and you have to enjoy that—not in a sick, weird way, just in the sense that this is what you are learning from. This is the way that it rolls, right?

Do you think you’ve become a better leader, respectfully, the richer you’ve got? The context I give for that, actually, is that I don’t worry so much, so I’m less emotionally volatile. I’m more neutral now because the downside’s a bit capped, and I see upside much more.

I don’t think it’s monetarily related, to be honest. You asked what Scott and I had in common. Neither of us are particularly monetarily motivated.

Right?

I used to love the old quote, “I didn’t become an entrepreneur to make money.” It doesn’t mean I can’t count. Every time we do those personality surveys and stuff, I remember we did one and some guy said, “You’re the weirdest founders ever.”

We were like, “Why is that?” He said, “Well, the importance of money to both of you is off the charts at the bottom.” So I don’t think monetarily, no. I do think, if I twist the question, the richness of my life has made me a better leader.

Having kids has made me far more empathetic to employees. In terms of all sorts of things—wisdom, pain, difficulty, all the suffering, whatever Jensen likes to say—going through all of those journeys, I think, as long as it doesn’t make you gnarled and hardened and instead makes you wise in a thoughtful way, that is what probably makes you a better leader, person, whatever. It’s the richness of the experiences that you’ve been through, rather than some sort of capped downside. I mean, I don’t have to look at the right-hand side of a menu. Great.

You said there about Jensen: Jensen said if he knew how hard it was, he would never have done it. My heart broke as an NVIDIA fanboy and a Jensen fanboy. It was like the shattering of Father Christmas not being real, which he obviously is for anyone listening who’s under the age of 10. If I asked you that, what would your response be?

Look, there are certainly times when I may have felt that way. If I took the average of all the days in the last 23 years, no, I’d do it again every single day. We’ve been—I still say “we” a lot. I have been, we have been, both of us, incredibly lucky. We’ve worked hard. We’ve got to work with amazing people.

I think, to your question on founders as well, I would add that, because I do think it’s really important: if you don’t enjoy the people you work with, none of the rest of it matters. I have an awesome executive team. I have some awesome, amazing colleagues at work that I’m inspired by, that I love working with, and that I love turning up to work with.

The old dinner-party conversation, I think, is really important. If there’s 1 seat left and you’re the last person to arrive, you go sit down. Do you look at the person on the left and right and think, “Ah, I don’t want to sit there”? Then you’re working with the wrong team. As a founder, it’s my job to make sure that there are no seats at the dinner table that I do not want to sit in.

At the same time, that’s where the joy comes from. What are we doing it for? A company is a group of people trying to achieve something in common, right? A team, a group—we have some sort of goal.

That's the fun part. Surely, it's like a sports team. It's like anything else: we're trying to achieve something, and at the end of the day, you achieve it. If your bank balance goes up or down, or something happens, those are kind of good monetary markers or external scoreboards, but as they say, you can't take it with you, right? You're going to enjoy the memories that you made.

So many of the old times at Atlassian, we talk about the hard times you've been through, the things you've built, the funny things you thought would work and didn't. You're like, "This is going to be amazing," and it totally flops. Other times, you're like, "Man, we didn't see this one coming." So that's all about the people you work with, right? All technology is built by people. And if you don't enjoy that, man, you're not going to enjoy the founder journey.

8. Quick-Fire Round

Dude, I've got to do a quick-fire round with you because I know that you have to get on with your day, too. But I've got to start with that: what's the biggest flop that you thought would be amazing that turned out not to be?

I don't know. A long time ago, we added status updates to Confluence. This was 20 years ago, when Twitter was all the rage, just because we thought it was going to be huge, and nobody understood it. It didn't work. It's probably not the biggest. I don't know.

What have you changed your mind on most in the last 12 months? For me, it was the magnitude of OpenAI and Anthropic. I didn't see them being 3- to 4-trillion-dollar companies in any world. And now I think, if you squint, you can.

Look, I think that's a good answer. I think they will both be very big companies. How big, I don't think anybody knows, but they will be big and important and impactful companies. That's probably true. Maybe, like many people, the speed with which AI can be understood and deployed successfully by companies and customers is something that's going to take a lot longer to really play through, because the delta between magical demos and actual value delivered is quite high in a lot of cases, and there's more work to do. Like most technical disruptions, we've got a hype cycle going on.

Why is that? Is that adoption and education? Is that data cleanliness and availability? Is that security and permissioning in large enterprises? Why?

Every single one of those and more. I think it's experience, it's understanding, it is people. You have to understand your problem. Well, you have to change the way you look at the problem—often the workflow, the business process. You have to understand the technology and what it is and is not capable of at the current time. You have to have the security, the data, you have to have the cleanliness, you have to have all these things.

The magical demo of just pushing a button and some app pops out, or this thing happens—it doesn't exist, right? It doesn't mean the technology is not useful at all, but I think there's a lot more work to do to really deliver on the promise. Like most technical disruptions, we've got a hype cycle going on.

My brother's just had a little baby girl, his first one. What advice would you give to a parent of a first-time newborn baby?

I would tell them that newborns are pretty hardy, right? So just relax a little bit. It always feels—it really reduces your world down to this tiny thing, and then you get incredibly stressed about it. I have 4 kids. They're all amazing human beings. They're all vastly different. They're all a total pain in my ass sometimes, and other times they're the most amazing people in the world. I wouldn't give them back for anything.

So just enjoy it. Try to appreciate the ups and downs, and don't stress about the little things. It really does reduce your entire visibility to this one little thing. They're pretty amazing creatures.

Do you believe your greatest strength is your greatest weakness?

Yeah, probably. I think it's really true.

What would yours be?

Probably being unreasonable.

It's full circle. I think almost always your strengths are your weaknesses. They are almost always two sides of the same coin, right? If you're optimistic, you can be naive. If you're unreasonable, you can be difficult. If you're blindly loyal, you can make it very hard to move on from things. They're always there. It's all about understanding the nuance and subtlety in the middle.

Penultimate one: you've got an "oh shit" moment. Who's your first call?

Scott. If I'm stuck in jail somewhere, it's probably still Scott. Very different problem at that stage. If one of my kids is having a problem, it's probably my ex, I guess. So it's very different depending on the context, but, yes, if it's work-related, it's probably Scott—if it's a real "oh shit" moment.

If we paint a picture for 10 years away, which is Atlassian 2035, where's Atlassian then?

Look, I hope we're still fighting. We're still creating. We're still—I hate the American term, but we're still a player, right? We're still fighting for customers, trying to build new things. If we're still doing all that stuff, we've done the right things, right? If we're still able to attract great talent and still doing all those things, we've done the right things.

I'm not sure if, when you say, "What do I want to be remembered for?" I'm even sure if Atlassian's it. I probably shouldn't say that. It's probably being a good dad. I don't know. There are probably 4 people I really want to remember me in the long arc of time.

But sure, it'd be nice if they thought Atlassian was a great place to work and we'd done some good things from Australia and put a little dent in the world. But on a personal level, look, at some stage, Atlassian has to live on without either Scott or me, right? I love what I do. I'm not going anywhere at the moment, right? I'm really enjoying things. But will I be here in 50 years' time? Probably not. I probably won't even be on the planet in 50 years' time, right?

So if Atlassian lives on and is competing, you don't want your company to live on as some sort of weird zombie. You know what I mean? You want it to live on as a vibrant, competitive place that is trying to deliver some value, solve some problems, create things. If it's living in that state, then I'm pretty happy. I think you look back and say, "Cool, it's still going."

Mike, I'd heard so many great things. As I said, Myles Grimshaw, Rich Wong, Scott, Heather Fernandez—dude, I stalked the shit out of you.

Really? Right. Okay.

But I so enjoyed this. Thank you for being so open and for putting up with some very wayward questions. You've been fantastic.

No problem. Happy to do it anytime.