There’s been a surge in AI investment but is a trillion-dollar spend by big tech really justified? Milford Investment Analyst James Buchanan talks with Ryan Bridge about the rise of AI agents, the early signs of commercial returns, the risks of an AI bubble, and what this fast-moving technology could mean for businesses, jobs and investors.

Click here to download the MP3 file or listen to the podcast on your favourite platform:

Bridge talks Business: 9 June 2026

Episode Transcript

Ryan Bridge

Kia ora and welcome to episode 79 of Bridge talks Business with Milford. The big US tech companies are on track to spend more than a trillion US dollars on AI next year, and a lot of that is being driven by the rise of so-called AI agents. This week we catch up with James Buchanan, Investment Analyst at Milford, and ask if this investment is justified. First, something different, here’s your top five AI trends.

    1. AI agents become genuinely autonomous. The biggest leap was the move from chatbots to AI agents that can plan, reason, they can take actions across software tools and workflows. Instead of simply answering prompts, models began using browsers, using apps, completing multi-step tasks, writing debugging code, operating software, a whole bunch of stuff more than you and I could probably do in a day.
    2. Multimodal AI went mainstream. Over the past year, AI has stopped being primarily text-based. At the beginning of 2026, frontier models could natively process text, images, audio, videos, PDF, even screenshots. Breakthroughs included real-time voice interaction, screen understanding, visual reasoning, and live translation. Pretty cool.
    3. Reasoning models became dramatically smarter. Basically, models got better at maths, scientific logic, coding, planning, structured problem solving. This means improved reliability and improved memory over long tasks. This was the year AI moved from impressive autocomplete towards something like junior knowledge worker.
    4. Open Source AI nearly caught the frontier labs. One of the most important industry shifts was how quickly open models closed the gap with OpenAI, Google, and Anthropic. Models from Deepseek, Alibaba, and other open-way projects became competitive on stuff like reasoning, coding, and multimodal benchmarks, often at a fraction of the cost.
    5. AI entered the physical world. This past year, we have seen major progress in robotic manipulation, emboldened and embodied AI warehouse automation, household robotics, AI-powered industrial systems. Instead of rigid programming, robots are increasingly learning through observation and reinforcement learning. In other words, they’re becoming a little bit more like us.

Alright, it is time for our feature interview and this week is a bit of an AI special for you. We’re asking the question, is a trillion dollars investment in AI justified or not? And joining us is James Buchanan, Investment Analyst at Milford. Just a reminder, this segment is informational only and should not be considered financial advice. James, welcome to the podcast.

James Buchanan
Thanks for having me on.

Ryan Bridge
Great to have you here. So a trillion dollars. I don’t even know how much money a trillion dollars is. It just sounds like a lot.

James Buchanan
It’s a massive number. So to try and frame that, it’s like five times the GDP of New Zealand. So like a big, big number. And is it justified? It’s kind of the big question we’ve been kicking around in the market for the last 12 to 18 months. And the market’s really been oscillating between, okay, are we walking into this AI bubble? Or is all this spend going to be justified? Are we going to see the returns? And I think we’ve actually seen the first proof points of that start to come through in the Q1 results. And Google was like the real standout highlight that we saw. So like a year ago, that cloud business was growing like less than 30 percent. It just grew 63 percent. And that’s going to be like a hundred billion dollar run-rate business by the end of the year. So it’s big numbers on big numbers and just lots more big numbers. So at the moment, like I would say we’re starting to see those returns come through. We’re seeing the revenues accelerate at those big mega-cap companies. And then also we’re seeing the margins expand at the same time. So we’re starting to see there are some justifications for all this AI spend.

Ryan Bridge
Right. So justification for the spend, not justification for the talk of a bubble.

James Buchanan
Yeah. So I mean, the other big question in the market is really, is this an AI bubble? And the real answer is that we never know if this is a bubble or not. But we can we can look at history and there are some analogues and we can do some comparisons. So, if we go back to 2001 dot com bubble, so everyone knows about that one. Those companies were not supported by real earnings – like Cisco that used to trade on 200 times earnings. So that means it would take you 200 years to get your money back, if you put a dollar in. You had companies like Pets.com who would spend 30 million dollars a year on advertising. And they only had revenues of five hundred thousand dollars. So there was just all this spend and there was nothing behind it. The difference this time is that we’ve actually got companies on reasonable valuation. So Microsoft, who was exposed to this theme, they’re trading on like 20 times earnings. So it’s a much more reasonable valuation and on no measure are we stretched yet. And then also we are seeing revenue start to accelerate. And so the difference with the dot com bubble is we didn’t have those tangible revenues to back it up. And this time we do.

Ryan Bridge
And the uptake is only increasing, right? More and more people are starting to use AI, businesses are, you know, employing it – in some cases in place of workers. You mentioned agents. What exactly is an AI agent and how important is that to the story that you’re telling here?

 James Buchanan
Yeah, so I would say this has been like the really big change that we’ve seen in the last few months. So if we go back six months ago, ChatGPT – it could do some wonderful things for you. But you had to drive everything. You were sitting there and you had to give it all the instructions. And so now we’re moving into what’s called the agentic era in AI agents. And this is where they will take actions on your behalf. So, for example, I guess you probably have to go around the country and do various interviews and someone has to organise your travel, your hotel, book the flights. ChatGPT could six months ago craft you an itinerary for that. But someone still had to go take that action and actually do the booking.  With an AI agent, you can now outsource all of that and it can know your preferences. It can know what hotel you like. It can know what you like on your pillow at night. And so it can actually communicate with the hotel. It can do the booking for your rental car. It can take all those actions. So there’s been a real material change. And so why that’s driving returns is that all these AI agents will be running around and they’re using compute. And so that’s resources and then companies have to pay for that. And that’s what’s driving the returns. And you can think of it as it’s going to be a digital employee for a business that never sleeps.

Ryan Bridge
Do you use it for those purposes?

James Buchanan
Yeah, I’ve got a few agents running around. I can give you a personal example, actually. So Mother’s Day. Big day of the year. Really important. And so I wanted to do something special for mum. I wanted to go out for a brunch, find a nice spot. ChatGPT – last year, it could go find me some nice brunch spots that we could go to. This year, it could actually take the action and it could book that. But Sunday, May, not sure if it’s going to be nice weather in Auckland. I wanted to do something outside, but wanted to have something to hedge my bets. So I got my AI agent to go and find two options that we could use. One that was going to be outdoors, one indoors. It went and made the booking. And then it said, do you want me to ask you on Sunday morning what the weather is? And if I should cancel the other booking. So it gets to seven o’clock on Sunday morning, open the curtains. Glorious day. AI agent says, how’s the weather? Great day. Can you go cancel that other booking, please? We ended up going to a place outside. Fantastic day.

Ryan Bridge
Okay, fascinating. But why couldn’t the AI agent look at the Met Service? Why does it have to consult you? Are there still tweaks that can be made to this?

James Buchanan
Yeah, so for next year, this is going to be great because then I can hook it up to Met Service and then I won’t actually have to message it. It will know already. I hadn’t actually thought of that one, that’ll be the next step.

Ryan Bridge
But companies, clearly companies are spending a lot of money on AI. What exactly are they getting AI to do at this point? And is it threatening jobs, you know, as we sit here today?

James Buchanan
Yeah, so if we think about what companies are doing with AI agents in particular, I think Aviva is quite a good example in the UK. So it’s the UK’s largest insurer. They’re now saying that 90 percent of their inquiries from customers are handled by an AI agent. And so, the average person in the street probably doesn’t know there’s all these agents running around doing all these tasks because it’s not actually really visible. And this is only going to evolve further. So Aviva, that’s going to move next into claims. And so if you were to crash a car like my wife did the other week actually. I had to call up my insurance company. I was on the phone for 45 minutes going back and forward. Roll this out in a year or two’s time, and that could potentially be all done in the background. I wouldn’t have to have any interaction with any person. And I could go straight to the panel beaters and get the car fixed. And so this is where we’re going to see companies are more in the testing phase. And it’s not very visible. But as we roll through the year, it’s going to really start to proliferate. We’re going to see a lot more of this.

Ryan Bridge
And are those jobs that the AI bot is taking in place of a worker? There was Eric Schmidt formerly of Google speaking to a bunch of students at Arizona State in the last couple of weeks, and he was talking about AI and all the student graduates were booing, because they clearly don’t like the idea. They think it’s going to take their jobs away from them. Are we seeing that yet?

James Buchanan
Yeah, so I think your job is probably safe. Mine is question mark on that one, I think. Because I can see that AI is doing parts of my job that I have spent 10, 15 years training to do. But what I found personally is the more mundane parts of my job can now be automated. And I’ve got more time to go and research companies. So I’m finding it to be a fantastic addition. If we look more broadly on the actual numbers, like what is happening, what companies are doing, developers are a great example where they’re right in the crosshairs of AI. Their view is that AI is writing all the codes. So you’re not going to need any developers. And so you would expect that companies will be cutting developers and you would see job listings for developers going down. But what’s actually happening is the listing for job developers are going up because companies can code up a lot more. They can go after new markets. There’s new products to build. There’s new opportunities to go for. And so this is actually a potential revenue opportunity for companies. And you need people to do that.

Ryan Bridge
Okay, interesting. What about the number of companies – or traditionally a handful of companies who have benefited from this? Who are the real winners? Is it still that handful of companies or is it starting to broaden out?

James Buchanan
It is. So at the moment, it really is concentrated around big tech. So those companies who have built the models or those who are selling the chips like Nvidia to the big cloud companies like Google, those have been the real winners and that’s what the stock market has priced. So we’ve seen semiconductor stocks, since the March lows, up 60 percent already. Like no concerns about the Iran war there. And so it’s been really concentrated there. But I think now we’re at the point where it will start to broaden out because the value will accrue across the entire ecosystem. So it’s moving from that first layer of the chips – the picks and shovels we like to call it – moving to the cloud companies. So that’s the hyperscalers – your Google, your Amazons. And then there’s going to be the potential for investors to look at the large language model companies themselves. So we’ve got big IPOs coming this year. We’ve got Anthropic, which is rumoured to come later in the year. OpenAI as well. They might come potentially Q4 or Q1 next year. So, there’s going to be more opportunities to invest in these companies. And then for actual businesses themselves, it’s a bit more nuanced because the first step that companies need to take is actually you need to invest in AI. So the first thing is it does cost you more before the revenue opportunity comes in or before you’re able to take cost out. So in terms of it getting into the broader economy, it will probably take a bit more time.

Ryan Bridge
I had to return to the bubble question, but the journalist in me just wants to go to steer towards the negative. How do you know you’re not in a bubble? Is it like a fish doesn’t know it’s in water kind of thing? Or I mean, granted what you said about Q1 and the results that we’re starting to see, would we actually know if we’re in one?

James Buchanan
We wouldn’t. It’s in hindsight – and hindsight is a beautiful thing. We don’t know until we look back. So that’s the really difficult question. At the moment we can only really deal in probabilities and then looking at some of those historical examples. And I guess signals that we would be looking for to signal that we are in a bubble, would be really, really stretched valuations, expectations that are far, far too high. And at the moment, we’ve just got the inverse of that. Valuations are not stretched by most metrics. And then in terms of the expectations, companies are actually beating expectations on revenue and earnings rather than actually missing those lofty expectations.

Ryan Bridge
Final question. Did your mum care that you used AI to pick the cafe?

James Buchanan
She’s got no idea. But if she watches this, I’m going to be caught out. And then, yeah, I don’t know how I’m going to resolve that next year.

Ryan Bridge
James, good to have you on the podcast. Thanks for being with me.

James Buchanan
Thanks a lot, Ryan.

Ryan Bridge
That was James Buchanan, Investment Analyst at Milford, talking to us about AI, the spend, the hype and the justification for some of the valuations that we’re seeing of these massive tech companies. It’s always great to have you watching and listening. Don’t forget you can like, follow and subscribe wherever you like to listen. Until next week, don’t forget, invest in yourselves.

Missed previous episode? Don’t worry! Click here to catch up now.