Elon’s bid for OpenAI, 5G Breakthroughs and Quantum Leaps

Episode Transcript (computer-generated)

Paul Spain:
Hey, folks, greetings and welcome along to the New Zealand Tech Podcast. I’m your host, Paul Spain. And fantastic to have Finn Hogan joining us in the studio. How are you, Finn?

Finn Hogan:
I’m very well, Paul. It’s a pleasure to be back in the studio. Always love talking nerd stuff with you, so thanks for having me.

Paul Spain:
Yeah, no, it’s great. Of course, we used to talk techy stuff quite a bit when you were at TV3 Newshub. I don’t know, the varying iterations as it came under international ownership and so on.

Finn Hogan:
Exactly, yes.

Paul Spain:
Tell everyone what you’re up to these days.

Finn Hogan:
Yeah, so I was tech reporter at Newshub for many years and you were actually putting it quite politely. I think what really happened was I would annoy you, constantly calling you at like 4pm being like, Paul, please, please, please, can I come put a camera in your face? I’ve got a story at 6 and I need you to do some commentary. So, yeah, I think you’ll be quite pleased that I’ve now since left that job and I’m now doing a bit of a split role. I spend half my time as head writer for Caffeine Daily, which is sort of a publication focused on the startup and tech ecosystem in New Zealand. And then I also do some work with DigPR, so I do some comms work around, specialising with New Zealand tech businesses, but have a fairly broad remit. So, for example, we do a bit of work with OpenStar down in Wellington, the astonishing fusion startup there. And yes, obviously people will write off what I say because I am employed by them as a client. But genuinely, anyone look up OpenStar and I challenge you to not be astonished that that’s the kind of work we do here.

Paul Spain:
Yeah, it’s pretty cool. I do look forward to delving into their story at some stage in the future, so we’ll probably have a separate conversation about that one. But before we jump in, of course a big thank you to our show partners to One NZ two, Degrees Spark, HP and Gorilla Technology. It’s their support that keeps New Zealand Tech Podcast ticking along over all these years. Now, let’s jump in. There’s a chunk to sort of chat about from a New Zealand perspective and then things on an international front and fair bit to cover, so let’s see how we go. But to start with Tū Ātea , who were previously for a period known as the Mori Spectrum Working Group, number of years ago they were issued some mobile spectrum, I think, what is it, 100 MHz? And they are now working With Centreport, which is the port in Wellington to deploy for the first time in New Zealand a commercial private 5G network. And this really caught my attention when I saw the email come through in the earlier sort of hours this morning because, well, I don’t think we’ve ever had a private mobile network in New Zealand before.

Paul Spain:
And so they’re talking about a private 4G, 5G network. And this was one of the, I guess one of the stories we were told when, when 5G was being touted as this is the new thing that’s coming. And these are all the possibilities. One of the examples that, that I remember, you know, hearing about and seeing a sort of probably a video walkthrough of was a factory floor where previously they’d had to run ethernet cables to all these machines. And that was the only way to get sort of fast enough and reliable enough communications. And you know, WI fi wasn’t, you know, quite right for, for the needs and so on, but that 5G was, was going to tick that box and so really fascinating to see there being a New Zealand use case. So yeah, kind of curious how it plays out. Tū Ātea they are in an interesting position because they have the spectrum and then through the previous acquisition that they did, I think a couple of years ago, they have quite a technical capability through their, what used to be Broadtech group, they acquired late 2023.

Paul Spain:
So yeah, I guess that’s a bit over a year ago. So, you know, they’ve got north of 100 staff and you know, quite interesting to see this come together now. I’m kind of curious how the, I guess the behind the scenes, the technicalities, what are the advantages of doing this with a private network versus say, you know, some sort of partnership with one of the existing telcos, what are the, you know, what are the pros and cons? And I guess these are the sorts of things that we might, well, you know, hear about in the future. But no doubt, you know, there’s been a chunk of work done and this has been decided as the, you know, the way to deliver, you know, the most exceptional communication capabilities and therefore, you know, helping them to increase, you know, automation and efficiency across the port.

Finn Hogan:
Well, yeah, exactly. I think there’s obviously the big picture considerations around taking, you know, a part of that spectrum private. And as you say, I think that’s probably the first time that we would have done that. But I do think it’s a fascinating for as you mentioned, well, if you’re doing a walkthrough of a Factory and there’s ethernet cables everywhere. I think a port is almost like the distillation of logistics. Right. It is the pain point, it is the choke point for all logistical supply chains. And so I’m really interested to see how this tech and its ability to split out and remove some of that kind of infrastructure that you have to put down to be able to smooth out, round out those edges on cargo and freight coming through, because What, Centrepoint’s like 4th biggest port by weight, I think, coming in.

Finn Hogan:
So I think it’s a really interesting use case because once we see this installed, I think the differences probably aren’t going to be super overt to someone wandering through. Right. But if you shave off even a fraction of a percent of time when you’re dealing with that kind of volume of freight, any tiny incremental gains you make across the port are going to start stacking up super, super quickly. And it seems like the 5G infrastructure in this very limited, very controlled and therefore very specialised way could end up having really profound effects over time because they’re going to sort of stack together and I think when they show sort of the proof in the pudding and they show some of the results they get from this, I think people will be watching this very closely because this seems to be the kind of thing that you could roll out in other areas.

Paul Spain:
Yeah, look, I am really, really curious and how do the dollars and cents sort of add up? How important is it that, that, you know, to Atiya have, have that, that spectrum available and yeah, is this just sort of the beginning of we’ll see a whole lot, you know, a whole lot more? So yeah, it’s, it’s a, it’s a really interesting one to, to have happening in, in New Zealand and of course we like, we like seeing new things happen with technology. So yeah, fascinating time ahead, the third quarter of 2025 is when they’re expecting to be launching that network. And yeah, as I say, we’ll keep a watch. Now, I should just mention also on the network connectivity front, and we’ve touched on this multiple times, but around Starlink and last week we had one NZ announce that they’d extended the devices out that were gonna work with their satellite to text capability to include iPhone. We were pretty sure that was probably imminent anyway because it had just been maybe a couple of days before our episode last week where Apple had released an update that was specifically mentioning support for Starlink for mobile to text type capability and so on. So yeah, sure enough, that came out interestingly and I don’t know if this is because I’ve got my iPhone set up to access kind of early versions or betas or some other thing, but when I went to try out the capability on the weekend, I pulled over whilst heading somewhere out of Auckland where there’s no coverage and in this particular spot I was able to see I was connecting to one NZ SpaceX cell site. So I was connecting. Wow.

Paul Spain:
I just didn’t have any luck with my texting.

Finn Hogan:
I was about to say, yeah, you were connected, but were you actually capable? That’s the difference.

Paul Spain:
So there was something there and I’ll probably try it again over the next week or two and I’ll probably. Maybe I should have had the Samsung phone with me as well and then I could have seen. Is this something with the phone software or was there something else going on? But I mean, it seems like their trial has been ticking along and of course most of the time we’re all in coverage areas so we don’t need that. But when you are out of coverage, very nice to have these capabilities. So I will report back on that one once I have success and yeah, that could be next week, so we will see how we go now. Also on a New Zealand front, Ivo, the AI startup that is set up to, I guess, assist with contract negotiations. They were known previously as Latch. They’ve secured a Series A funding round and US $16 million raised.

Paul Spain:
So this is, this is quite significant, isn’t it?

Finn Hogan:
Very much so. And I think we’ve talked before and it’s very much the theme of the year in these circles of agentic AI, is coming in 2025. And 2025 is going to be the year that we start seeing actual value returns from these AI systems which have previously been treated a bit more like a novelty. And, and I think this is a really great example of an area that’s so ripe for disruption because so much of low level law, and I don’t mean low level in a disparaging way, but that kind of routine that a lawyer goes through, Matt, doing contract law, something that is very formalised a lot of the time is essentially, you know, changing names in something that is a stock template. It seems very obvious that a very well trained, specifically trained AI system, particularly ones powered by the latest generation LLMs, will be capable of producing work that is up to a legal standard. Obviously no one’s going to be trusting it with something that’s immensely significant, sight unseen, but I think this, I’ve been waiting for someone to do this And I’m really glad that it’s happening in New Zealand that we’ve seen so much success. But I do think this is the start of something we’re going to see a lot more of. And once they show their success as well, once you’re entrusted, once that kind of levy breaks, once people start entrusting routine lore to a Gentec AI system, I think it’s going to really kind of break the ice a little bit and allow many more consumers to be comfortable with this, particularly if the price point’s right.

Paul Spain:
And we’ve also got Lawview in New Zealand, of course, operating in the space. So. Yeah, and that’s not the only one. There was another previous one whose name has popped out of my head, who chatted to in the past. So yeah, fair bit of movement here, I guess. Yeah, you can just sort of imagine how this could play out. Remember a friend of mine, I think he was, well, he was working in radio and he used to have an agent to kind of, you know, negotiate his, you know, his salary arrangement or however the contracts were. And it would go, you know, these things would go backwards and forwards.

Paul Spain:
And I imagine that was probably a reasonably expensive thing to do. So I’m kind of imagining this scenario where, you know, a job offer or a contract thing comes and there’s all sorts of scenarios obviously in which, you know, contract negotiation can look, there’s, you know, buying and selling businesses houses and you know, umpteen other things and negotiations between, you know, businesses on contracts. But yeah, it would be very interesting. We get to a time where, you know, no human touches these things and the offer goes from, you know, one agent, you know, AI agent back to the other one and backwards and forwards. And then, you know, afterwards you get a bit of a sort of a summary overview on this is what happened. And here’s where they’ve got an agreement in principle by the AIs. And now it’s down for, you know, the formal legal team to do a final review and a sign off by the chief executive or something.

Finn Hogan:
I can’t imagine this is going to be controversial. There are a lot of controversial use cases for AI. But I think anything like this, which is very formulaic but very expensive is going to be in that perfect kind of realm for AI disruption currently for the early stages of agentic AI. So yeah, I really can’t see this being controversial to people. I think it makes a lot of sense and I think when you see people like Arcanum AI and Wellington as well kind of doing a Similar thing around routine admin. You can really just start seeing the breadth of capability these systems have. And I think particularly when you see OpenAI dropping things like deep research a couple of weeks ago, I don’t think the whole year of AI agents thing is hype. I think we’re already seeing the capabilities and I think the first half of this year is going to be the market being introduced to some of these capable agents.

Finn Hogan:
And then I think the back half of the year is very much going to be those agents taking off.

Paul Spain:
Yeah, I’m still curious around how long it takes for these things to really work, because there are significant sort of shortcomings that we still see with AI. And yeah, it’s just like sometimes we imagine these things are going to move really, really quickly and sometimes they do, but if we don’t sort of solve the kind of gotchas in there, then that becomes quite a stumbling block.

Finn Hogan:
Right, I agree, but I do think that there is a mid path to walk. Like, obviously these things are still gonna hallucinate, no one’s gonna trust them implicitly. But I think what they’re getting good enough at now is collating and combining huge amounts of information very, very, very quickly and then presenting it to you in such a way that it’s very easy for you to go through and check. And so I just feel like there’s going to be at least a few hours saved out of the day for someone if they truly implement these systems. And if those few hours are from someone who is very specialised and is very capital intensive to deploy, I can’t see how that’s not going to be pretty economically disruptive.

Paul Spain:
Now, moving on, we’ve got a scenario here in New Zealand and this is. Mental Health Minister Matt Ducey has announced it. The Mental Health foundation is gonna receive funding from the government’s $10 million mental health and Addiction Community Sector Innovation Fund for what they’re calling Project Wakapuāwai. And this, I guess the initiative aims to digitise mental health resources to, you know, improve access, make them more available, particularly for those in rural and remote communities. And so, yeah, apparently they’re going to the foundation. The Mental Health foundation will match the funding supporting development of a digital, you know, platform that’s gonna offer a range of wellbeing resources, capabilities to help with suicide prevention and information to help those sort of suffering from mental distress. So, yeah, really interesting to see this leaning in on technology to really help improve where things are at from a mental health perspective.

Finn Hogan:
Yeah, absolutely. I mean, I think that there isn’t any more worthy application of funding than trying to help people’s particularly in mental health distress. Right. But I do kind of want to, I’m curious about the details of exactly how that will be implemented. I think what happens sometimes with these kind of implementations is they say we’re going to make resources more accessible, but that’s a fairly broad remit. Right? Like what kind of resources are going to be available? Because a lot of the time what people really need is to talk to someone, they need to talk to a human 100%. And I think what often happens is sometimes we’ll say, oh well, we’ve got a website that’s really nicely redesigned and the UX is really good and you can access and read through all of these mental health sources. And I’m not disparaging that as a part of the system, but what I hope is that this isn’t implemented and what we actually get is a nice refresh and a very slickly designed interface.

Finn Hogan:
But what really needs to happen is when people really need to talk to someone or they need to access acute care, that’s possible, but of course that ends up being a lot more expensive.

Paul Spain:
There has been some interesting work that’s gone on with AI and chatbots within this area and it’s kind of encouraging to see some of the data that’s come back suggesting that those types of mechanisms can be quite useful. Of course, if you integrate that into a broader mechanism and then allow those sort of triggers of, hey, it sounds like you’re really going through a super tough time. Can we connect you with somebody by phone or can we book you an appointment? So a more integrated approach there could be really helpful. Of course, what you ultimately want is that we bring down sort of the instances and the things that lead to some of these sorts of challenges. But, you know, we can dream of nirvana, but these things don’t necessarily happen overnight. Now also one other thing from a New Zealand perspective where I see there’s an opportunity for technology to really come into play. And I think, you know, most of us do, but you know, we’ve heard the announcement of New Zealand’s new emissions kind of targets to reduce emissions by 51 to 55% below 2005 levels. And this is now by 2035.

Paul Spain:
So there’s been a bit of rejigging of targets and things as we to a degree expect, as, you know, different governments make different commitments and by bumping it out to 2035, you know, there’s all sorts of things you could Debate around will we ever get to the dates that we’ve committed to, or will that then get scrapped and a new date? And there’s all sorts of things floating around. I saw there, there was something from Federated Farmers saying, well, you know, these targets are gonna cost $24 billion to reach. I haven’t really found a lot of depth and substance behind that. And when we were chatting about all of this before the show, there’s so much data and sort of stats lying around, but usually when you start thinking, oh, I’d like to see it sliced and diced this way, you can’t necessarily actually find it.

Finn Hogan:
I mean, having worked as a political reporter for quite a few years and then as a tech reporter, there is nothing more labour intensive than reporting on climate stories because exactly as you say, the numbers are so dense. And people from the left and from the right love to throw around big numbers. This is going to cost X or climate devastation is gonna be Y. But it’s famously one of the most difficult things to predict at the top level. From a climate system perspective, it’s one of the most complicated systems ever. And from the business side of it, when we start projecting out what costs are, I think when you really drill down on those, there is always such a wide range of possibility. So, I mean, without even kind of trying to get into the numbers to circle back to the tech side, one thing I would say is that this government has been pretty consistent at deflecting criticism for sort of softening some of our targets by saying, well, we’re going to be leaning into tech. We’re gonna be making sure that we’re taking advantage of emerging technologies to help us reach those targets.

Finn Hogan:
And that’s something that always just twangs in my ear when I hear it, because that’s essentially the go to thing for any government trying to reach its climate targets is, well, don’t worry, tech’s gonna save us. And I’m a tech optimist. You know, I wouldn’t be here if I wasn’t. But every time I hear that, it just reminds me of a smoker being like, don’t worry, I won’t quit’cause they’re gonna cure cancer. You know, And I do want to sort of see some specifics around. Exactly what are you doing to help us harness that emerging technology? Because let’s face it, over this government’s tenure, there hasn’t been a massive amount of support for the tech ecosystem. The massive restructure of what’s happening in R and D, this Establishment of Callaghan, you know, the Marsden grants. We’ve seen a lot, sort of, at the very least you could say restructured and not a lot added when it comes to actual investment.

Finn Hogan:
We’ve famously lagged behind our peers when it comes to R and D. We’ve never really bumped up those numbers. So for me, I kind of want them to put their money where their mouth is when it comes to. Okay, if we’re going to depend on tech, how are you making that happen?

Paul Spain:
Yep. And what do you reckon? Fusion?

Finn Hogan:
What do I reckon about fusion? Yeah. Is that going to be the thing that saves us?

Paul Spain:
That could be part of the picture.

Finn Hogan:
I think fusion is a fascinating part of the puzzle and I think there is such renewed interest in nuclear over the past sort of decade as we’ve started to see the really acute dangers of climate start to be kind of thrust in everyone’s faces. And there’s this interesting story happening now of as AI’s power needs just start storming up. I mean, I think I saw one stat of data center power consumption in the US around 2018 was something about 2%, but it’s projected to reach up to 10% of the grid within about a decade. Like the needs of AI are so enormous. And that’s why all of these tech titans are shoveling billions of dollars towards new age nuclear projects, including a lot of fusion. Sam Altman’s heavily invested in fusion projects. So there is an idea that if we want to unlock the kind of AI future that we’re all dreaming of, that kind of Star Trek future, we first have to get a new nuclear age to power it. But of course, and again, I am very much interested in fusion and I’m pro the technology.

Finn Hogan:
The critics do have a very strong case for saying this will never be commercially viable on the timelines needed to make a difference to climate. Because famously building any kind of nuclear, let alone an unproven technology like fusion, takes a decade longer and billions more dollars than you ever expect. And we need to do things right now.

Paul Spain:
Yeah. And yeah, would New Zealand have the appetite for it as well? Is another aspect, but we’ll maybe pick that up another time. So jumping into sort of international news. OpenAI, is it up for sale? What’s going on? Elon Musk and supporting consortium apparently waving around a virtual $1997 billion US offer, or if we put that in New Zealand dollar terms at the moment, around $172 billion. So an eye watering amount. Now this seems to be below the Overall valuation for OpenAI in fact, I read something today that said their latest funding round could put OpenAI. And this is a kind of, I can’t remember whether it was New York Times or who was reporting this one, but they were talking about potential $300 billion valuation. So.

Paul Spain:
Yeah, Rather fascinating to sort of watch this going on while Musk’s doing all his other stuff with US aid and you know, getting stuck into sort of the, the politics there in the, in the U.S. yeah, he’s already got a fair amount of stuff on, on his plate. But this is, this is the latest, you know, part, part of, part of the puzzle.

Finn Hogan:
Yeah, him and Sam Altman’s bromance really turned toxic and we’re all just paying for it. Cause, you know, it is astonishing, but people may have forgotten and they kind of the onslaught of Elon Musk news. But, you know, he founded OpenAI with Sam. They did this together. And this in a lot of ways sowed the seeds for this drama that we’re seeing today. Because Elon Musk has always been a pessimist around the dangers of AI. He’s always said it’s one of the great threats. And at least in public, what he’s saying now is OpenAI has moved away from its original mission of making artificial general intelligence safely.

Finn Hogan:
But I think what’s interesting about this is it kind of gets at this interesting moment for OpenAI where Sam Altman is attempting a very delicate thing of unspooling the non profit arm of OpenAI and spinning that out from the company and then having a completely for profit company. Because up until recently, because with DeepSeek, which we get into later, we thought the only way to make these models was with an insane amount of money, hundreds of billions of dollars. So that’s where that valuation comes from. But that was announced before Deep SEQ really shook up the market. These guys are looking for hundreds of billions of dollars. And so some analysts are going to look at that going, well, hang on, that’s an insane amount of money if it’s possible to make these models so much cheaper. So that’s kind of will change the framing around the entire ecosystem, which maybe we could talk about a little bit later on. But to come back to Musk and Sam Altman, I think what’s interesting here is that the OpenAI has got such a complicated structure.

Finn Hogan:
While the board is under no obligation to just take the offer that comes to them, they do have a fiduciary obligation still. And that non for profit arm still controls OpenAI at the moment. Remember the board, they’re the ones who got rid of Sam Altman for five days. Obviously a lot of those left. He’s now stacked it with loyalists. But if Elon Musk comes to them and puts a giant offer on the table and they reject it, they still could open themselves up to some legal challenges. Well, you’ve got an obligation to return value on this company, even on the non for profit arm. So I think even if it doesn’t get accepted, which I don’t think it will, Elon Musk has kind of thrown the gauntlet down at a really, really inconvenient moment for Sam Altman because Sam Altman’s trying to unspool this, but that part of the company that he’s unspooling didn’t have a valuation previously.

Finn Hogan:
But now Elon Musk has come out saying, I think it’s worth $97 billion. And he said they’ll match any other offer. So the board will have to consider that of like, okay, if we get another offer on the table, how high can we get this number? And then that’s gonna be a nightmare for Altman. But I did love his response on the platform formerly known as Twitter, where he said, replied to Elon Musk saying, no thanks, but I will buy Twitter for $9.7 billion. Just moving the decimal place of his offer over by one. And then Elon Musk responding, being like swindler, like, it’s so petty. It’s like Real Housewives of Silicon Valley, these guys. And it would just be purely funny if the potential fate of the world and the world’s most important technology weren’t at stake.

Paul Spain:
Yes. Yeah, it is a very interesting time in history to potentially the final time in history, but there’s a lot going on here and yeah, pretty hard to kind of get your head around the whole picture. And I guess, yeah, you’re looking back on OpenAI setting up as this, this nonprofit. And of course, you know, Musk was, was right there alongside Altman and others at that, that founding stage. And yeah, now here, here they are, yeah, battling it out. So, yeah, it will be fascinating to, to keep watching this journey, I’m, I’m.

Finn Hogan:
Sure because there’s going to be a Oscar winning movie made of this. The social network part 3, completely made by AI but I’ll love to watch it.

Paul Spain:
A lot more twists and turns ahead, I’m, I’m sure. Now Google are saying that they expect to have commercial quantum, quantum computing applications within five years. Whole, you know, range of things, material Science, drug discovery, energy solutions, as, as quantum computing advances, this is compared with Nvidia, sort of suggesting this is maybe more like a 20 year journey ahead. Now, that’s a pretty significant difference. And obviously we’ve been talking about quantum computing for a long time. This next level of computing which just completely takes the lid off from a performance perspective and this ability to do within seconds or minutes computing tasks that previously you would look at as things that might take years or decades. So, yeah, very, very interesting to kind of see this heating up a little bit. And I’m curious as to whether this is kind of spin from Google.

Paul Spain:
Are they, you know, are they genuinely confident? I mean, the amount of tech stories that we hear in all sorts of areas from, oh, there’s a new battery technology that’s going to be in cars in, you know, five years. I mean, I remember probably, you know, discussing a story like that on TV3. Yeah. Probably over a decade ago. Oh, these new batteries that are going to. And it was like, oh, that’s amazing. You know, 10 years down the track, we’ve had a small percentage increase in battery capacity in electric vehicles and a few changes in the technologies, but nothing as dramatic. And of course, quantum computing has been something we’ve been on the verge of for quite a few years.

Paul Spain:
What’s your take, Finn?

Finn Hogan:
Well, I think it’s interesting. Quantum computing, fusion, AI, all these technologies we’re discussing, we’re in that realm of science fiction and we’re always just over the horizon and then at the same time are starting to advance very, very quickly. And I think they feed into each other. Right, because quantum computing is also seen as a way that we will reach the kind of AI future that we’re seeing. The kind of compute that a quantum computer could bring to training an AI is just a next order of magnitude. But I think with the quantum question, it’s a yes, but. Right. I think it’s not just spin.

Finn Hogan:
This is Google we’re talking about. These are one of the most powerful, most brilliant companies that’s ever existed and they did have some data to back themselves. It wasn’t just saying. When they released their chip last year, they showed that it had a lower rate of decay on the qubits because, I mean, again, I do not have the degree necessary to fully break down how quantum computing works. But essentially, instead of zeros and ones, which is the binary part of any computer, a quantum computer can do multiple bits of math simultaneously. So 0, 1 and 0 and 1 at the same time because of those weird properties in quantum mechanics, where something can be in a superposition. And I think the things that they talked about with their quantum chip coming out last year were astonishing and they were real, but the context around them is important. So when their big headline grabbing feature was they solved a math problem using their quantum chip that would have taken a traditional computer the entire age of the universe to solve, actually much greater than the entire age of the universe by many orders of magnitude.

Finn Hogan:
But when you drill down on that, what really happened was they basically worked out the perfect math problem to pose to a quantum computer, because they do operate differently than traditional computers and they’re very, very, very specialised at certain things. So it was essentially, how do we engineer the perfect problem for this to solve, but that quantum computer can’t do 99.9% of the things your traditional computer can do. And I think there’s also a bunch of more mundane practical concerns of. Currently, the only way to get a quantum computer to work is to get it to basically absolute zero. You have to cool it down to a fraction of a degree above absolute zero, because any heat introduced into the kind of molecular system that makes up these chips confuses the qubits to the point where they’re not. They can’t make any kind of accurate readings from them. So the logistical side of quantum computing has a long way to go before it’s commercially viable. And the actual use cases, even though they’re broadly discussed, haven’t actually been specifically defined in the sense of how exactly is this quantum computer going to be more generally useful than something that I can get at an absolute fraction of the cost.

Finn Hogan:
So, yeah, I do think there are really interesting applications down the road. But when they say commercial applications, do you mean maybe if we have a huge amount of money from government, we’ll have a couple in the country?

Paul Spain:
Yeah. And look, it’ll be governments that really are most likely to be in the position to want to leverage these. Although I can, you know, the commercial cases I’m sure will be there as well in time, for sure. Now, deepsea, we’ve talked about this before. We’re now in a position where. What’s the list we’re looking at? South Korea, Australia, Italy and Taiwan have banned Deepseek, which is the Chinese AI system, on government devices, due to concerns around security and data privacy. We’re still seeing other countries, such as New Zealand and the United States, most likely doing some considerations behind the scenes and working out whether they should follow in those footsteps. So, I mean, when we look at deepsea, everything that we’ve heard about looks incredible in terms of what it can do.

Paul Spain:
Small amount of computing power. You know, they built this thing supposedly for next to nothing, or there’s been some debate around, you know, what, what the realities are there. But we seem to keep coming back to issues with Chinese companies when it comes to data and privacy. So, you know, TikTok’s of course, in that picture as well. How do you see this playing out? I mean, from my perspective, I think DeepSeq potentially is a really key part of the landscape from an AI perspective in the years ahead, particularly with open source versions of their technology. But then we got that kind of intermingled with a whole lot of situations where it’s banned and a whole lot of use cases. That becomes fairly complex.

Finn Hogan:
Yeah, I mean, I think there’s sort of two points with Deep Sec at the top level. There’s the. It’s getting caught up in the geopolitical question, which is the. And like you mentioned, it’s essentially being pushed in the same way that TikTok was. Because we’ve already established the TikTok ban, it would almost be difficult for countries to not investigate banning it because, to show the consistency, because at least TikTok had a plausible deniability of saying, well, no, we’re separated out and yes, at the parent level, but with deepseek, absolutely not. There’s no degree of separation. So if you’re concerned about privacy, of course you will ban it. However, I do think the company can push back pretty credibly, saying, well, if you want to use our system and you’re concerned about access to it, obviously they’re not going to admit anything like this, but they can say, just download it yourself and inject any of your own biases into it.

Finn Hogan:
We are being much more open and transparent than OpenAI is with the way their system works. Here’s a very detailed paper of exactly how we trained it. We are providing you so much more clarity on how this works. And if you don’t like the answers it gives you, run your own local version and make it give you the answers that you want. So I do think just on that level, there’s some interesting pushback they can make. But I do think in the broader ecosystem, deepseat could be really, really consequential because as we sort of touched on earlier, if they’ve proven, and it seems like they have, no one has really truly challenged their numbers in any significant, significant way. It seems to be that they have trained a model that is as good if not better on certain dimensions for radically less money. If that’s true, it really challenges the fundamental assumption that this AI bubble, or at least boom, has been built on, which is the only way to make advanced AI models is with incredibly expensive chips chained together and hundreds and hundreds of billions of dollars.

Finn Hogan:
If that’s not the case, suddenly there’s going to be so much capital floating around that might not just be harvested by three tech trillionaires. And I think there is a really optimistic case here that what deepseek shows is that there’s actually a much lower barrier to entry to this. And if that’s the case, who’s really going to benefit is startups, is companies like the ones we’ve just talked about on this podcast today. If you can think of an interesting niche application for AI, suddenly these investors are looking around being like, well, why am I throwing $300 billion at OpenAI when I can split up that bet into much smaller investments and spread that around? And then maybe AI truly will be one of those foundational technologies like the transistor, like electricity, where no one can truly capture the value of them, and the value instead seeps through the whole economy and just raises the standards that general customers can expect. And I really hope that’s the future. And it genuinely might be the case that Deep SEQ is the first step towards that.

Paul Spain:
Yeah, look, if all of these things are true that have been posited around deepseek, then, yeah, it could be really a key part of that, because the initial suggestion, what was it, US, $6 million worth of computing power to train it versus north of 100 million for OpenAI sort of GPT4. So. So, yeah, this is a really huge difference how it plays out in terms of how easy is it for you to. Or for organisations to sort of set up and run their own versions, and how good is the code base and secure and private and all those sorts of things? No doubt there’s some work ahead on that front, but it does make me wonder, putting that together with Musk’s kind of bid for OpenAI is, you know, surely those involved in that consortium would have looked at things pretty closely. And if deepseek sort of potential and disruption is at the level that’s been suggested, then you question what would be the value of OpenAI? Would it be anywhere near as high as those sort of figures? And that’s kind of interesting to get your head around, but we will leave that for those with a lot of money in their pockets to maybe explain to us through their Investments over the weeks and months ahead.

Finn Hogan:
Yes. But I will say I do think 300 billion. If that valuation ever truly eventuates, that will be the peak. I think it’s safe to say that. Well, I don’t. It’s never safe predicting anything in the AI world. But to be one of the most valuable companies on Earth in the kind of environment that we are in, AI right now, I do think that will be overvalued.

Paul Spain:
Yeah. Now, Doge, the Department of Government Efficiency.

Finn Hogan:
We can’t get away from Musk, can we?

Paul Spain:
I mean, if we look back over the years, I mean, Musk just keeps turning up in stories left, right and, and center. And look there, there’s been a, you know, a bunch of coverage saying, hey, you know what, you know, what Musk is, is, is trying to do through, through Doge in terms of, with, with US Government entities is matching what he tried to, or what, what he did with, with Twitter, with X wants to come in and, you know, cut all these jobs. And I find it quite fascinating because you’ve got these sort of, you know, two narratives sort of coming through and on one side it’s like, this is terrible, you know, you know, we’ve got lives at risk and, you know, a whole range of sort of negative consequences. And this is largely down sort of political lines, I think. So on one side you’ve got that and that sort of makes a lot of sense. On the other side you’ve got, hey, there’s a whole lot of fat and government’s wasting all of our money. And actually, you know, we should do this. But there’s this kind of Silicon Valley esque kind of pace at which it’s being done because, yeah, it is appropriate that governments should be audited and they should spend money in appropriate ways.

Paul Spain:
But I find it pretty hard to kind of see through the messaging on both sides. I’m not seeing a kind of a good overview that’s really weighing these things up kind of in a clear sense. What’s your take on what’s going on? Do you think that Musk is just completely nuts or do you think actually if we look back in, let’s say the end of the next four year cycle, who knows how long this thing’s going to take? That actually, on balance, people look at it and go, oh, that largely worked out okay, or are people going to be looking at it going, man, a lot of lives were lost and it was just a real shambles. And, you know, the US Economy isn’t any better off.

Finn Hogan:
I think Personally, I think it’s a nightmare on a few dimensions, but not for. Yes, of course, there’s always kind of hysterical discussion around anything in government or anything involving Elon Musk. But I think just on the raw facts of the case, the access that he is being given is inevitably gonna be challenged in court. Already we’re already seeing the Legal Challenges Department.

Paul Spain:
Right.

Finn Hogan:
And famously, it’s kind of Constitution 101 that Congress gets to determine the actual spending, and the president has much more limited power than Congress to access the kind of systems that he’s accessing. You’re having these crossed streams of the three branches of government. And Musk is obviously, he’s just doing the move fast and break things, I’ll see you in court mentality that he’s taken in Silicon Valley because he knows that if he moves quickly, he will do the things that he wants to do. And remember, one of the things he has done is demand access to the software system which controls $7 trillion of federal spending. So that’s also. That’s going to international agencies that’s dealing with a vast amount of the overall spending of the US and is accessing not just the data of the people that this is going to, but potentially being able to challenge that money reaching those people in the first place. And it has been frozen. There’s been the most messy messaging around this possible, like you mentioned.

Finn Hogan:
But I think whatever you think of Musk, you. I think the important thing to hold onto is we should not be giving an unelected person this kind of power. Yes, governments should always have to be held to account for how they spend. But every government does that. There’s almost every Republican government and a lot of Democratic ones will have some kind of board of efficiency. It’s not like this new thing that he just invented of, hey, you know, the government sometimes spends too much money. There’s been a million versions of this. But the fact that he has been given this kind of power and he’s been let off the leash so much by Trump, and he just seems to be expecting, I’m just going to race ahead.

Finn Hogan:
I’m just going to do these things and then they can, and then I’ll just get challenged in court. I don’t care. I can fight any legal case, and court cases famously move slowly. He’ll be in court cases about this for the next five years, but it doesn’t matter. He’ll have done the damage at that point. And importantly, as the Republicans love to say, once you give power to anyone involved in A government, they will never want to give it back. Cause they were always the hey, we want small government. We want small government.

Paul Spain:
Yeah, yeah.

Finn Hogan:
If try and imagine a world where it was Joe Biden in charge and he went to Bill Gates. You just go wherever you want in government and demand that any kind of services get shut down. Here’s access to $7 trillion or George Soros, whoever. Pick your billionaire. The Republicans would be in conniptions. They’d be having heart attacks in the House. And there’s a reason that Donald Trump is doing anything. All of his actions so far have been through executive order.

Finn Hogan:
Cause he knows he can’t get any legislation passed if he actually took it to Congress. So for me, I think Musk is brilliant in so many ways, but I think in this case he’s trying to have that move fast, break things mentality from Silicon Valley. I’m gonna run the country like I would run a company. Well, the closest analog would probably be Twitter. And look what happened to Twitter. It’s cratering. It’s not worth what he paid for it by a mile. Users are down, advertisers are fleeing the platform.

Finn Hogan:
If that’s your use case.

Paul Spain:
Well, that’s an interesting one. Cause I’ll, you know, I’ll pull you up on that. Because I was looking at this previously because that’s the messaging I always hear and I always, again, you see sort of two sides. One side says it’s doing better, the other side says it’s kind of cratering their stats, sort of suggesting that it’s. Yes, yeah, that, that it, that it’s growing from, from the X side. But I saw some figures the other day and, and I don’t know, I can’t recall where these came from, but I’m going to, in this case, give a bit of benefit of the doubt and mention them. But basically they were talking around the revenue that Twitter was bringing in was roughly, and this is pulling a little bit out of a hat here, but roughly say half what it was before Musk got involved. But the difference was that they were making a lot more profit.

Paul Spain:
So actually, from a business perspective, yes, the revenue is down, but it’s actually become a profitable business now. And they’ve indicated that time on the platform and users and so on is up now. I know there’s a fair bit of debate, there’s a lot of people that have left the platform as well. But when I look at it for kind of tech content, I’m still finding the same sort of stuff that I found before. So I think, you know, there are multiple aspects to this. And I guess when I look at sort of tech companies and the Silicon Valley approach, which, you know, has been incredibly disruptive in the business world and, you know, we’ve seen so many companies disrupted by the Silicon Valley startups. And Musk companies have obviously been, you know, amongst those that have done well often. That’s actually, you know, it’s quite exciting what those entities can do.

Paul Spain:
Now. What I don’t know is, and this is the kind of, hey, where we’ll be in two, three, four years, can that thinking this, you know, tech startup approach apply in government and deliver good results, or are we gonna kind of see, say, the worst that we’ve seen from tech companies? And I say, look at Uber, where, you know, Travis Kalanik had Uber moving at this super fast pace, coming into markets, ignoring regulations and what was legal to kind of dominate and get its work done. And then in some cases, like in Asia, you know, they’re no longer. They’re no longer there largely. And yeah, a lot of, you know, people and probably entities have been, you know, negatively impacted because they’ve kind of just ignored the rules and regulations and jumped in and blown things up. So, yeah, there’s that kind of genuine kind of, you know, curiosity and I think it is quite hard to, yeah. Remove it from. Yeah, just all of the opinions about, you know, about Musk and how he operates and about how Twitter has or hasn’t gone over the last two or three years.

Finn Hogan:
Totally, yeah. And look, we could go back and forth on the numbers all day. I only thing I would say is I do think running a government is fundamentally different than running a company. And I think often people like to conflate those things. And it’s a very easy talking point for someone saying it’s just common sense. I’m gonna come in here and I’m gonna run a government like I run the company. It’s like, well, no, because we’ve got democracy, because the government controls money supply. Like, the differences are infinite and the stakes are so much higher.

Finn Hogan:
And so I just think while it’s easy to watch from the sidelines and be astonished and be entertained, I think because everything Musk does is entertaining. Love him or hate him, I think the consequences for this could be so, so high. And I think accountability has to be brought into it. This person was not elected to this position and he’s not accountable to anyone because Donald Trump has yet to call him out on anything because really, Trump needs him as much as he needs Trump because those guys are so enmeshed, they’re such a unit now he’s the world’s wealthiest man. So yeah, I will always push back on the kind of move fast, break things coming to government once the nuclear codes are involved. Let’s just not move fast and break things, you know.

Paul Spain:
Yeah, well there’s probably some issues there, right, with how old some of that legacy technology is around those, you know, those sorts of things. And look. Yeah, and I do, you know, hear it around, you know, non elected and so on. But of course, and every government appoints people to these differing roles, but it does seem to be a role that is very unique. We’ve never seen anything like it before. So yeah, we’ll wait and see. Fortunately, it’s happening in America, which is a reasonable distance away for some of us. And I know some of our listeners are more US Silicon Valley domiciled, so they might be disproportionately impacted.

Paul Spain:
But as always with politics, these things tend to swing one way and then swing the other way and sort of, you know, move around. So, you know, I try not to get sort of, you know, too deeply concerned because, you know, that it goes too far one way, it’ll probably flip back the other way.

Finn Hogan:
Oh yeah, absolutely.

Paul Spain:
And so on. Right. So yeah, but certainly this is a time to watch. Now one thing I touched on briefly last week was this new HP Omnibook Ultra flip that HP sent across and had a bit of a play with this laptop. Fairly grunty little laptop, small and light, but why I was quite keen to have a look at it has got the second generation of the Intel Core Ultra processors. So we’ve gone from our, yeah, I guess what did we get to 10th, 11th, 12th, 13th, maybe 14th generation Intel Core. Now we’ve moved on to this new series, the Core Ultra and those have sort of been coming through alongside the ARM processors and hp, you know, like the Lenovo’s and so on, I guess you could say having a bob each way. So they’ve got the ARM based Windows laptops now, which are pretty impressive in terms of speed and battery life.

Paul Spain:
What we’re seeing with the Core Ultra, this sort of second generation and this one is an i7, is that the battery life of the intel chips is heading towards that of the ARM chips. Now we’re kind of used to the ARM chips in our phones. MacBooks with the M1, M2, M3, M4, you know, being super efficient and just delivering great battery life. I’ve also been, you know, trying out one of the ARM based Microsoft Surface laptops over, over the last few months and have definitely been seeing that battery life improvement maybe not quite as far as as the Mac. And yeah, I just wanted to have a, have a little bit of a dabble and, and get a feel for where are intel kind of going with these new chips. And sure enough, definitely an improvement in battery life. But I think they’ve, you know, they’ve certainly still got a bit of a journey ahead of them to sort of deliver that, you know, amazing battery life alongside, you know, being able to run everything that runs on an intel chip. And there certainly still are some sort of shortcomings when it comes to the ARM based processors on Windows.

Paul Spain:
So I think for individuals that are just using it for home use and the Omnibook Ultra Flip is a, you know, it runs Windows home so it’s not designed as a business device. But this is what we kind of get used to. The consumer devices come through quicker and then the commercial devices will come through afterwards. But it’s pretty encouraging and I think we’re, we’re moving into a period where yeah, those who do spend a lot more time away from their desks are going to find ultimately good Windows machines delivering the sort of battery life that the newest MacBooks have been delivering over the past few years. And yeah, that’s an area where Apple really jumped ahead. So yeah, thanks HP for sending that across. And of course what we’re also seeing is this sort of built in AI neural processing unit. So there’s that local capability.

Paul Spain:
What I’m not seeing yet for the typical average user is a whole lot of functionality that really leverages that. And that’s kind of, I guess, part of, yeah, part of this time we’re in, we’re starting to see these things come through, whether it’s in our Apple phones, Samsung phones now, HP and other laptops. Is this capability there for doing local AI processing? I don’t see it being leveraged to maybe the degree that hopefully we will in the future because I see some benefit and I guess keeping some of what we do with AI completely secure in our local device and not having it all thrown out to the deep seats or the Microsofts or the OpenAI’s. You know, actually there could be some benefit in keeping a bit more of that local 100%.

Finn Hogan:
Well, as power goes up, the concerns ramp up at exactly the same pace. Right. So as it gets more capable you will be more concerned about your privacy. So I think you’re absolutely right. I think Being able to locally run these things is going to be crucial.

Paul Spain:
Now before we finish up, just maybe just give folks who aren’t familiar with Caffeine Daily a bit of a summary, some of the stories and things you’ve been writing about recently.

Finn Hogan:
Head over to our substack would be the easiest way to find out. We’ve got a newsletter that comes out every day around 9 o’clock which sort of gives a collation and summary of some of the most interesting stories in the sort of startup ecosystem. I’ve got a column that comes out every Friday about sort of my rants and thoughts as you’ve just got a taste of here. But yeah, just head to Caffeine Daily on substack if you want to have a look. Yeah, we’d love to have you and yeah, it’s a very good time. We get basically a little in long form written format what we just did here.

Paul Spain:
Yeah, yeah, no, really, really helpful and look, I think it’s an important part of our tech ecosystem here in New Zealand and a welcome addition that we’ve seen in recent years to the mix. I think the first few years of New Zealand Tech Podcasts we kind of keep looking around, it’s like, oh, that media outlet’s disappeared and that kind of happened with most of the sort of long running tech media. So yeah, really, really pleasing to have Caffeine Daily providing that tech fix on an ongoing. Basis. So yeah, good job, keep it up.

Finn Hogan:
Thank you very much.

Paul Spain:
Excellent. All right, well thanks everyone for joining us on this episode of the New Zealand Tech Podcast. And of course a big thank you to our show partners, Gorilla Technology, One NZ, HP Spark and 2degrees. And we’ll look forward to catching you again next week. Week. If you’re listening to the audio, then track us down and follow us on the varying video platforms. YouTube X and LinkedIn are probably the main ones and I think we probably got Facebook in there as well. So yeah, thanks everyone.

Paul Spain:
We’ll catch you again next week. See ya.