In this interview, Paul Andon, Dean of UNSW Business School, interviews David Tudehope, CEO and co-founder of Macquarie Technology Group. David co-founded Macquarie Technology Group in 1992 and has guided the company's development from a provider of voice services to a fully integrated carrier and highly successful ASX-listed company specialising in telecommunications, cloud services, cybersecurity and data centres. David is also a highly esteemed Alumnus of UNSW Business School and a member of the UNSW Business School Advisory Council.

Following is an edited AI-generated transcript of their interview.

Paul Andon: Hello everyone, I'm Paul Andon, Dean at UNSW Business School. It's my great pleasure to be here today with David Tudehope, Chief Executive of Macquarie Technology Group. Thank you, David, for your time.

David Tudehope: Good, thank you, Paul.

Paul Andon: If you could start off by getting from you a brief description of your business journey, and I guess as part of that, a sense of the key elements that have guided you along the way, because one thing that you've been quite successful with your company is evolving alongside technological innovations that have emerged over time. So, I'd be very fascinated to hear how you've been able to construct that journey for your organisation.

David Tudehope: Well, thank you, Paul. So, yes, I was one of your undergraduate students back in the late 80s, and it was a fantastic part of my life. I've got a soft spot for the University of New South Wales, and particularly the Business School from those days, and it's so good to see the way the Business School has just gone from strength to strength since that time. For me, after I graduated, back then, the bank and financial services sectors were deregulating, so it was the place to be. My first career was in financial services. Then I started Macquarie Technology Group and went from what used to be called a small business back then (nowadays it's called a tech startup), which sounds a lot more fun. And we built the business literally from a very small number of customers, together with my co-founder and brother, Aiden (who's also one of your undergraduate students), to where it is today. To your question about change, though, the one thing that has characterised our journey has been changing technology, and the need to constantly adapt, innovate, and pivot has been, throughout the journey, one of the keys to our success. It’s something that you see; so many tech companies that don't keep adapting, and there are so many tech companies that go from roosters to feather dusters.

Paul Andon: And how do you keep on top of what's happening in terms of trends? You're saying you need to be agile and adaptive, but you also need to make the right bets. Have there been ways in which you've been able to keep on top of what's happening, so that you do make the right bets? And from your history, you've seen that you've made some pretty good bets over time.

David Tudehope: You have to constantly look up and look at what's ahead and see where the opportunities are. You constantly have to have the high beam lights on, as well as the headlights and the parking lights, and not just identify those trends, that's super as a starting point. I see so many companies in our sector that either don't see what's ahead, because usually it's visible to many people, but they don't make the decisions. Or, in some cases, they choose every option, which, of course, dilutes their efforts and effectively amounts to choosing none. So, for us, the way we approach that is what we call the Noah's Ark. What that means is for the last 30 years, we identified typically between five and maybe eight people, we deep dive on one or two big topics every few months. We found that the diversity of people in skill set and role type is actually where you get the best insights, combined with an intensity of focus on just a narrow set of questions that allows you to really come to grips with what technology is relevant in Australia (which often it's not). Often, there are many tech trends that never leave the US or larger markets, for various reasons: regulatory, market structure, and more. At the end of the day, it's a choice. And I'm proud to say that a number of those things we've looked at, we've actually decided not to pursue. I think that's always a great test of whether you, even after all that work, have the courage to say, ‘it's a great opportunity, but not for us.’

Paul Andon: The process sounds fascinating. I'm sure it is really important and useful, and very much part of your success in the way that things have developed at your business. Can I ask about the latest foray that the company has had into data centres, and we're sitting in one of your facilities now. It's an impressive facility that continues to grow. From your point of view, what role do you see data centres playing in society, and in the Australian society and economy, more specifically, as we are moving into a more AI-enabled world? And how might that even contribute to thinking about national economies and related issues?

David Tudehope: To your point, data centres are where the cloud, AI and cyber live; it's their home. So having that foundational home in Australia makes complete sense, and it's part of the value chain that Australia can participate in, and has been participating in, which is excellent, and something it needs to do more of. I think the other opportunity is, of course, what you can do further up the value chain, and that's equally important. The other part of our business involves our own cloud services, supporting foreign cloud services in Australia, supporting software that runs on top of those services, and supporting AI and cybersecurity. So all of that is critically important, but it all lives in a physical data centre. Another important aspect of data centres is that they enable data to be stored in Australia, and I think that's something that should be of great interest to your alumni and the broader business community.

Paul Andon: Yes, data sovereignty is a topic that's often discussed. Your views on that, in terms of what we should be doing as a society around building greater capability around managing and maintaining our own data on our own shores?

David Tudehope: I think people often just click ‘I accept’ when they're buying software, or they're buying a service, but in the ‘I accept’, quite often, it says the data can be kept anywhere at the provider's discretion. What people don't appreciate is that this means the data can literally be stored anywhere they choose, including a foreign jurisdiction. I think there's a very well-established principle in law that, if I use the airline analogy, you could be in an American aircraft, you could be an Australian citizen, but if an issue happens over Indian airspace, you're actually subject to Indian regulations, which means their courts, the Indian regulator. And you go, but how could it be? It just happened over their airspace. It's the same with data: if it's kept in an offshore country, like, say, India, you are now part of the Indian regime, which might be in your interest. But it might be a complete shock to you as well, which means Indian law applies, the Indian regulator applies, and also all the privacy rules are now Indian equally. If it's kept in China, it's now China, or anywhere else, for that matter. So that may matter to your business, and it may also matter even if you're not a regulated business or a critical infrastructure business; it might matter one day in a way you hadn't fully foreseen. An example of this is privacy. Australia has excellent privacy protections and is a global leader in that space. A lot of thought goes into it. Other countries choose not to do that, and I think a lot of people would be quite surprised to learn that all those privacy safeguards they thought were in place don't apply to them if it's been kept offshore.

Paul Andon: One thing you've made a big bet on is that data centres should be centralised in a sense, having big data centres that centralise activity across a range of organisations and other needs of other entities. There are benefits to that, obviously, as opposed to, for example, a university trying to create its own data centre on campus. What are the benefits that you see from organising that way?

David Tudehope: This has been a long-term trend for 25 years. We built the first data centre in Sydney 25 years ago, and back then, there was a lot of angst from IT managers saying we needed to keep it on-site; this is critical for our business. What happens if it doesn't work? The world's moved on significantly from that, partly through generational change, partly through just education, have recognised that having a computer room, or sometimes they call it a data centre, inside the corporate office or the university has a number of significant drawbacks. I mean, yes, it's nice to be able to pop in and wander around, but the power inefficiency is significant. The reality is, building data centres at an industrial scale with modern equipment is super powerful and efficient. If you don’t do this, it is also less reliable, because unless you have unlimited budgets to build a data centre for a few hundred million dollars, like we and others in the industry, your data centre is not going to be reliable, because you've spent a few million on it, or even a few hundred thousand, and there's a trade-off to that. So, reliability is a big issue. And finally, having multiple carriers, and in our case, we've got over 20 different external carriers. This is important because connectivity is essential. Having your services somewhere but not connecting to anything defeats the purpose. So that combination suggests a trend away from private computer rooms. It's very rare now for people to build new data centres or computer rooms in their offices. But there are examples, remarkably, and I think there's a well-established trend for all those reasons.

Paul Andon: If I can shift gears now a little bit and talk a little bit about what's on everybody's minds around technology, and that's artificial intelligence, and what that's doing to not only technology, but to workforces, to society, etc. Just a very general question to start with: from your perspective, do you talk with many big organisations that are looking to implement AI? A lot of those clients are probably using your data centres to host that data. What are you seeing, and what are you hearing in terms of where organisations are trying to take their businesses in what is fast becoming more of an AI-enabled world?

David Tudehope: I think everyone knows AI is the revolution of our lifetime. I'd say it's probably equivalent for those who can recall back to the internet in 1998. There are some things you can foresee today, like in 1998, well, I don't think we’ll have encyclopedias going forward. But there are a lot of things you could not foresee in 1998 as to where internet would head, I think it's similar now, people are focusing on certain very obvious consequences, they're seeing some incredible personal productivity gains, and people saying it’s transformed my personal life, my workplace life. It's because it's natural language text, I don't need to be a coder to play in this space, that's all true. And of course, it's just going to get better and better. I think there's quite a disparity right now, and it's not necessarily size-related. There are some size-related issues, around the use of AI outside of personal productivity, in terms of work processes, in terms of reasoning, in terms of more systemic sort of company-wide initiatives. While large companies have probably had teams, they're bound to put on projects to create some early examples, early proofs-of-concept, as we call them. I would not bet that a large company today is going to be the leader in its industry. I think this is an incredibly exciting opportunity for business owners, people working in up-and-coming businesses, whether they're like true startups or the kind of late-stage or maybe well-established businesses, to completely turn the tables on their larger competitors, because the amount of legacy systems and legacy processes involves a huge amount of work. Equally, I think for large organisations that do embrace the change and are willing to make all those disruptive things, that's so much harder in a big business. So, this is an exciting time. It's actually a fantastic time to be in every industry, because AI is not about the tech sector; it's about everyone.

Paul Andon: Yes, and I think the excitement really captures me, too. I think there are concerns about things like the ethics around AI, or the way in which AI might produce or exacerbate biases in decisions and other activities. But I think the opportunity is a really interesting aspect of what businesses can do to move forward in what's going to be a radically different space. You mentioned the productivity ceiling; if you like, everyone's jumped in. They're all using it for their own purposes. They're improving their emails; they're increasing their productivity in very individual ways. But the big leap, and I guess the big gap at the moment, and this is something that OpenAI is talking about, Anthropic is talking about, is the leap from personal productivity to actually realising that. They talk about it as a theoretical potential of AI: how AI can do more than just those personalised things and actually help transform workflows, helping to really bake in productivity across a workforce. Can you explain what you think that might look like? You talked about how some smaller organisations might leap ahead on that, while other larger organisations may be too wrapped up in their legacy systems. How do you see that barrier being broken? Is it just that there are organisations out there that are hungrier and have less legacy and other constraints that keep them back, or are there other things you think will also contribute to how we move forward, just from gains in personal productivity?

David Tudehope: Paul, there are so many interesting dimensions that you've touched on. Maybe I'll just start with your opening comment: I do think in Australia we have some excellent ideas for creating some guardrails, ethical business, legal guardrails. But I think the balance is far too much on the side of governance and not enough on the path of embracing opportunities for our country. And I think you see that, not even under the Trump administration, but back in the Biden administration, the bias was toward opportunity, and the Biden administration had some guardrails that were still forming but had a sensible basis. Even around in Europe now, which I think is a very, in my opinion, sophisticated approach to many issues in my industry. The European Union, I think, is one of the leading regulators in the world in so many places now. Even there, you've got to lean much more toward the opportunity than toward the governance. You have both, but people are saying, “What is the opportunity to transform?” So, I think that's just maybe to your first point. Back to the more practical pieces, I think the biggest delta is data. And when you move from personal productivity to business-wide transformation, the question is: How do I leverage the data I have? These are insights into my customers, these thousands, millions, tens of millions of records I have of interacting with my customers and interacting with equipment over the last few years, a few decades. How do I use that data, which I've always considered to be intellectual property, as a competitive advantage? How do I use all of that, and the people's knowledge as well, and utilise that in the world of AI? 

For a corporation, that's a lot of competitive advantage in that information and in the people that sit on top of it. I think that is a really interesting topic. Now, the simplistic one is, well, it's fine, we'll just expose the company's data set to the model, and then, well, off we go. Then, of course, that is effectively like publishing it to the web, so that doesn't work so well either. Happily, there are ways that you can manage that now, but the first starting point is, ‘How do I make sure my data is sufficiently clean?’ No data sets are completely clean, sufficiently clean that I don't get answers back that are nonsensical or with bias in the data in a way I hadn't fully appreciated. So, cleaning the data is a really big deal. I think a lot of the corporates we see are really embracing the opportunity, struggling with data cleaning, and sometimes trying to skip that step. Because it is hard work and, in many cases, has been neglected for years, the IT team hasn't been heard about the need to invest in cleaning the data. Well, now it really matters, and I think a lot of AI business cases have to do some foundational pieces before they can go straight to how do we turn that proof of concept that the CEO or the exec might have pioneered, into something that can be company-wide. And that's, I think, a really interesting topic, and one that we spent a lot of time with core technology, thinking about how we do that for our customers. We've got some great systems and processes there, a lot of which are driven by AI to clean up the data before an AI model can utilise it. The next part is: having done that, how do I create a cybersecurity governance framework on top of it? Because doing it once is great, but then, how do you keep it clean? How do you know whether, when we put new data into it, it doesn't pollute what you work so hard to clean?

And how do I protect security? It would be really embarrassing if all these data sets I've cleaned up and now have beautifully presented (with a few errors, maybe) were then hacked. That would be like one step forward and five steps back. So the security piece, managing the data, and getting it out are critically important to a client. Unsurprisingly, we've spent a lot of time working on developing solutions for our customers, but we think it's critically important to bridge that gap. And the final piece is, now we've done that, what AI model is right for us? There's a really interesting strategic battle right now between open-source models, which are fantastic, and more closed models, where they're trying to create lock-in and obviously extract an economic prize for it. Corporates are struggling with which way to go, and the answers, unfortunately, are not straightforward, so they have to choose one or the other. Intellectually, people go, well, that'd be open source one, so I'm not locked in, and it's inverted commas, “low cost”, but of course, there's an overhead of a different kind in managing your own model, and the skill set is obviously one of those overheads. So it's a really interesting question. Where do you go next, once your data is there, and you want to apply it to a model? The fantastic news is that there are lots of really good choices. The open-source models are incredible; the closed-source models and the frontier models are also incredible, and it's changing by the month. And even some of the projects we've done, we spent a lot of time working on one model, and got to the point where it's working really well, but a new AI model comes along, and we go, ‘wow, what we've just spent a year working on, we probably could have done in a month with the new model.’ So it's that much change.

Paul Andon: You're absolutely right, there is so much change going on, and everybody's just trying to figure out exactly what this looks like for themselves. Can I go back to your first point about the opportunity versus the governance, and to your comments about how we need to rebalance? Do you think that's particular to the Australian market, or are you seeing it play out similarly overseas? I'll tell you a very short, playful quip I've heard: the US innovates, the EU regulates, and Australia hesitates. Is that something that you think is right, or do you think that's a misconception? What do you think about how Australia is looking at its future with AI?

David Tudehope: I'm a big fan of everything Australian, as an Australian company, obviously. But sadly, that quip does, I think, strike a note. There is an element of that, without question, and I think there are some parts of that we should be proud of; it's great that we care about consumer privacy and respect people in the way we approach governance; this is a good thing. But the European Union, I think, shares very similar values at the moment, so it's not like you can't embrace that at the same time as seeing the opportunity. I would say it's more of a leadership question, and I think it's about how to provide that leadership at both the political and corporate levels. I think quite a lot of the people who often fill the media with comments can rebalance their comments to recognise the incredible opportunity this presents. So, I think it's a bit more of a leadership issue, Paul, than any specific action.

Paul Andon: I've just come back from a conference overseas, an education summit, where OpenAI was talking to a number of education leaders from around the world about what the future of education looks like with AI. One of the things they presented there, and it gets back to your issue about productivity, is that it's not only organisations that are finding it challenging to move from personal productivity to actual organisational gains; there's also variance across nations. The ones that seem to be moving ahead more are probably the ones that have the most to leapfrog or the most to gain, I suppose, whereas more advanced nations may have, like organisations, legacy systems, legacy issues, legacy legislation, and other things. Are you seeing, in your interactions with companies or other connections around the world, that countries looking to leapfrog are really leaning into the opportunity more? How do you see it playing out on the international stage?

David Tudehope: There are many lessons in that for Australia. I think there is a school of thought summed up in Australia, maybe it goes beyond hesitation, a sense that maybe we have too much humility in Australia, which is normally a virtue. But occasionally that can actually work against us, and there is no reason in the world where Australia can't be successful in the world of AI – not just from data centres, which is, I think, a really clear place where Australia can play in that part of the value chain – but also the development of both cloud platforms and AI models and cyber security. Some emerging countries are seen as an opportunity where they can create some comparative advantage, move early and hard; I think there's a wonderful lesson for us too, because I actually do believe that this is not a time for humility, it's not a place for us to go, ‘well, you know, we've missed the mark, US big tech is going to conquer the world. We're just going to be takers, we're just going to be the land of multinational salespeople selling their licensed platform,’ which often happens in the software industry. This is actually a place where we can be makers, not takers. I think, as Andrew Charlton called out, that definitely talks to your example. There's no reason we can't be like that, too. And it's not too late. The reality is that AI is in its very early days. I mean, my view is in 1998, nobody would have said, looking back now, but even back then, no one would have said, "Well, you know, it's all done by 99. Like you know, the reality is that in our tech sector, people can move very quickly. I've learned over 30 years from fantastic, successful models to struggling businesses, they can have a neon sign in North Sydney or in Melbourne over the Yarra, and the next moment they're gone, they've been merged and disappeared. It's happened so often in our sector, and I say that from that point of view, there's no reason why Australia can't be successful. Yes, the AI models Australia has may be niche, but we have some niche sectors where we're very large, and we have very deep capability. Frankly, one of them is research, and the other is also in the education space. We have clearly world-class institutions in education and research. There is no reason we can't also be world-class AI researchers.

Paul Andon: Can I ask you, to follow up on that, what do you think Australia should do now to sit at the forefront of AI-driven value creation and innovation, and as you've alluded to, not just simply be a consumer of AI technology?

David Tudehope: The first part we talked about is that leadership is a big part of it. The second part is: how do you create some elements of the value chain in Australia so these businesses can be successful? Obviously, data centres are one of those, but then there's all the bits in the middle. There's a need to ensure there is an opportunity for AI platforms in Australia, and I must say, credit this time to the federal government, which recently made a very large $200 million investment in our business for that. Well, one of the key purposes was: how do we build a model in Australia? The answer they saw was back in a business like Macquarie, and the government invested in us, and that's about creating a platform that, if you're an AI company, Australia, and you go, ‘I want to deploy to Australia because of the data sets, because my customers are here,’ or just frankly, ‘because I'm here, it's more convenient to grow a business.’ I can. I don't have to go offshore. That's super exciting, I think, and a super important element of the value chain that enables people in AI businesses. I think the final part would be the skill set, and I think there's a critical role there, for, of course, for universities, but not just universities, but all those organisations are attached to universities. They have incredible skill sets, and there's a lot of maths and STEM types at work there. Ultimately, it's about deep data analytics and working with large datasets, skills that many researchers take for granted. They spend their entire careers potentially working with large datasets in the medical sciences, business, and engineering. This is their life, and the opportunity I think for researchers to frankly be super successful tech co-founders or entrepreneurs, or maybe just generating research that is a breakthrough at a world scale – the kind of research that doesn't just simply fill an academic journal, but one that people actually refer to, the one that's quoted, the one that's crosses over into the broader community. This is the opportunity, and combining their deep database skill sets with new AI frontier models, whether open-sourced or closed, I think, is a chance for talented academics, medical researchers, or other researchers to punch well above their weight. Maybe they're in quite a modest institution by even Australian standards, but there's no reason they can't leave their sector.

Paul Andon: You mentioned a bit about education, and looking at that from a technological point of view, you're a Business School graduate from UNSW, and a longstanding and very esteemed alumnus of ours. You've also had the opportunity recently at one of your in-house conferences to talk to the 25-year-old version of yourself. If you were to talk again to your 25-year-old version of yourself, if he were going through university now, what advice would you give him in terms of becoming ready for the AI-enabled world that we are rapidly moving towards, and what, in a business degree, do you think would be important for him to focus on?

David Tudehope: Well, the first comment would be just from a study point of view. The great news is that I know nearly all your courses, and you've now got significant tech elements in them. It's no longer, well, ‘that's the computer science stream, take that course.’ So that's the first one. I think it has transformed a lot in recent years. I think the other part is to conduct more large-scale quantitative analysis, work with large databases and statistical tools, and develop software architecture. I think that's very valuable. The advice many people would have given 10 years ago was to do a coding course. There's no harm, of course, doing a Python course and getting some foundational pieces, but it doesn't actually matter that much anymore. There's a role for it, but it's not as critical as it was three years ago. There is much more in those other elements, so that's the university element, I think. Coming out of university, many people would say, as I did, that they wanted to start their own business. The advice I had back then, which I think is still right even with the past few years, was what my grandmother actually said to me: "David, you really want to work with someone else first and learn about business." I think she also said, “You want to make mistakes with someone else's money,” but I think it's also true. I know that's always a bit of a dampener, because they're often so enthusiastic about wanting to start their own business. I do think, as a young lady or young man graduating from the University of New South Wales, there is incredible value in spending some years, whether that's two years, three years, four years, maybe 10 years, 15 years, whatever it is, working for someone else to understand how businesses succeed and how businesses are structured. Technology is one thing, but they're part of something else. It's not just that technology lives in its own ecosystem. I do think there's incredible value to working for someone else. And it might be that you actually find that having your own business is incredibly hard. Even with modern technology and the need for the same level of capital investment up front, having done it, I'd say the level of sacrifice you’re required to make I think is unrealistic for many people. However long you think it's going to take, I always say it'll take three times as long. You think it'll take two years and that you'll be successful. Well, it will take six. Everyone goes, "Well, that's alright. I've got a supportive family or supportive partner, and it will be fine.” Well, in my case, I didn't go on holiday for the first eight years. And you go, "Oh, that's all right.” Well, I mean, like, zero holidays. They go, “okay”. Well, I could just stay around Melbourne, Sydney or Brisbane or something. You go, “Yeah, well, how about you don't eat out for the next six years?” Oh, well, that'd be different. Yeah, that's home cooking for the next six years. That's not Uber Eats; it's like home cooking because you haven't got any money. And I think when people realise the level of sacrifice to start your own business, that's just not for most people, and I respect that. The odds are no. People sometimes say it's 50%, but I think it's much lower: only about 50% of businesses actually succeed.

Accept the fact that you might have a one in three chance of being successful, and you might have to pivot a couple of times along the way, and you go, ‘wow, that's a lot of money, a lot of time, a lot of personal investment. That's not me. I'd love to be part of that, working for someone else, but that's not me personally.’ I think that dynamic is misunderstood, and people who are successful in businesses tend to skip over that bit, or there are some who are just lucky. They just land their feet, and miracles occur. But that's the odds of that are really small.

Paul Andon: Look, thank you so much, David. I really appreciate the time and the opportunity to talk, and thank you very much.

David Tudehope: Pleasure. Thank you.