Lukasz de Pourbaix (00:02)
Hello, my name is Lukasz de Pourbaix, Head of Strategic Sales and Solutions at Fidelity, and I'm joined by Co-Portfolio Manager for our Australian Equities Fund, Sam Heithersay, and Associate Director for Sustainable Investing, Sue Lyn Stubbs.
And today we're going to be unpacking a little bit around AI, which has obviously been a very topical theme. This one is going to be really interesting because a lot of the news we hear around AI is very much at the top-down level, whereas I think what’s interesting here is that you’ve had over 31 meetings with boards of ASX-listed companies, which is a very unique perspective to provide to our listeners. It really dives deeply into how they are navigating AI, how prepared their companies are, and where the opportunities and risks sit.
So with that, welcome Sam and Sue Lyn.
Sam Heithersay (00:55)
Thanks, Lukasz.
Sue Lyn Stubbs (00:56)
Thanks for having us.
SaaS sell-off and AI market sentiment
Lukasz de Pourbaix (02:01)
So one of the interesting things we observed late last year was a big sell-off of a lot of the software companies, and it was described in a number of different ways. One term—which I’m going to get wrong—was “SaaS-pocalypse,” I think. Anyway, you’ve probably got your own terms.
But effectively, we saw a big sell-off of software companies globally. One of the interesting things is that the ASX, the Australian market, actually sold off a lot more than the US market, which is counterintuitive when you think about the tech exposure within Australia versus the US.
So I’d be interested in your observations as to why that occurred, and your views on that sell-off.
Sam Heithersay (02:42)
Sure. I think you’re right in characterising the SaaS-pocalypse as fairly severe in the way it played out in the Australian market, and fairly indiscriminate.
I think it’s important to assess what happened because we are seeing a bit of a fight-back from some of these SaaS names, particularly in the US. The index there has been up 15% in the last five days. But people are also questioning the return on investment from a lot of this AI spend.
So it’s really important for us to assess where we’ve been in order to understand where we can go from here, and whether we should be buying at these levels.
You’re right in that it was quite severe and indiscriminate. To put some numbers around that, we saw the ASX 200 de-rate from September to March from 20 times to 17 times earnings. That was more than any other developed market, and it was really led by the tech sector, which had a de-rating of almost half—from about 100 times PE to 50 times PE.
Now, you could attribute some of that to a pivot in RBA expectations from cutting to hiking. You could also argue that Australia has more AI losers than winners, particularly compared to the hyperscalers.
But the general premise was that AI-native startups were going to disrupt the business models of software companies, and if they didn’t kill those businesses, they would replace the software coder on which many per-seat revenue models rely.
However, we saw such a severe sell-off that it raised some red flags, and from a bottom-up fundamental perspective, it didn’t make a lot of sense to us.
We don’t have as much tech here compared to offshore indices—it’s about 5% versus 30–35% in the US—and we’re also lagging the recovery. The Australian market is more dominated by materials and financials, which are more insulated from AI disruption.
And something Sue Lyn found as well is that trust and adoption in Australia seem to lag other developed markets.
Sue Lyn Stubbs (05:02)
Definitely. When you look at international surveys of AI uptake—whether that’s trust, adoption or implementation—you can see that Australia ranks quite low relative to other markets.
Many OECD markets are lower compared to global leaders, but Australia still ranks relatively low even against those peers. I think that was reflected in many of the conversations we had with boards. There was a degree of conservatism and hesitation—questions around how to implement AI, how fast to move, and what it means for the workforce.
That said, the debate has shifted somewhat. Boards are becoming a bit more open and more ready to implement AI and use it across the business from a value creation perspective. But overall, we’re still seeing hesitation and a very conservative approach from most Australian boards.
Lukasz de Pourbaix (05:57)
And just unpacking that sell-off a bit more—why do you think it was so brutal from an Australian perspective?
Sam Heithersay (06:05)
I think there are a number of reasons, some more perception than reality. The RBA pivot didn’t help, but there was also a broader perception that Australia has more AI losers than winners.
That was based on the assumption that the winners would be a narrow group—hyperscalers or frontier model developers—which Australia doesn’t really have.
However, if you look at AI disruption over the longer term, the second phase is more about diffusion and adoption. From that perspective, we do have a number of companies that could benefit, but the market wasn’t really pricing that in during the sell-off.
So for those reasons, combined with already lagging trust and adoption, it didn’t quite add up to us. That’s what prompted a lot of bottom-up work to identify winners and losers more clearly.
Governing in the Age of AI
Lukasz de Pourbaix (07:17)
From a governance perspective, Sue Lyn, you’ve published a paper called Governing in the Age of AI. What were your high-level observations from that work?
Sue Lyn Stubbs (07:45)
Overall, we identified a gap between implementation and governance and risk management.
When we spoke to boards, most of them are approaching AI from an efficiency, cost and productivity perspective, rather than from a revenue generation or customer growth perspective.
So we’re seeing a bit of a split between how CEOs talk about AI—focusing more on revenue growth—and how boards describe it, which is much more conservative.
We also observed differences in where companies are in their journey—whether they’re experimenting with pilots or rolling out broader implementation—and questions around what this means for their business models.
AI governance and risk
Lukasz de Pourbaix (08:46)
Given the level of access you have, did anything surprise you in those meetings?
Sue Lyn Stubbs (09:09)
One thing that stood out was how AI risk is being managed. Many companies are integrating AI into their existing enterprise risk frameworks rather than creating a standalone AI risk framework.
That’s notable because AI tends to amplify existing risks. It’s something we’re continuing to focus on—how companies define and manage AI-specific risks, including perception risk.
Another key issue is “shadow AI”—employees using AI tools on personal devices. This raises potential IP leakage, reputational, and legal risks. It’s something that isn’t being widely discussed at board level, but could be material.
Identifying AI winners and losers
Lukasz de Pourbaix (10:24)
Sam, from a stock perspective, how has this shaped your thinking?
Sam Heithersay (10:46)
It was crucial in helping us build a framework to assess AI winners and losers across the ASX 200.
We developed a set of characteristics to evaluate companies, and had analysts apply that framework to their sectors. That helped us identify potential mispricing opportunities.
The framework considers things like digital vulnerability, task commoditisation, and the strength of competitive moats—whether through proprietary data, regulation, or integration into customer workflows.
It’s important to overlay that with the risk that these moats may unwind. As Sue Lyn highlighted, we’re still in the early days of how companies assess and integrate AI risk, so something like a proprietary data advantage could be vulnerable—for example, through an AI-enabled cyber attack.
So we combined governance insights with bottom-up analysis to assess each company and identify where there may be mispricing, and ultimately where we see potential winners and losers.
Adaptability as the differentiator
Lukasz de Pourbaix (12:33)
When you think about winners and losers, it’s such a fast-moving space. AI is evolving rapidly, so separating winners from losers can be challenging.
Do you have any examples investors can anchor to when thinking about this theme?
Sam Heithersay (13:03)
I think it’s important to distinguish between a winner or loser today and a company’s ability to adapt over time.
Looking at past disruptive technologies is helpful. Take Kodak and Fujifilm in the early 1990s. When digital photography emerged, Kodak was unwilling to cannibalise its core business, whereas Fujifilm pivoted into healthcare and survived.
If you’d analysed both companies at the time, they may have looked similar in terms of risk. The differentiator was adaptability.
So when we think about winners and losers, we focus on companies that can adapt and pivot. The best indicator we have of that is governance—how mature a company’s thinking is around AI. That gives us insight into their ability to navigate disruption and make better long-term investment decisions.
Sue Lyn Stubbs (14:40)
Yes, absolutely. Governance can act as an early signal of which companies have the right culture to adopt AI, as well as the structures to manage the risks it creates.
We’ve seen cases where companies cite a very large number of AI initiatives—100-plus use cases, hundreds of pages of documentation—and from an investor perspective, that can actually raise concerns about capital allocation and returns.
So governance helps us assess whether companies are focused on strategic value creation, rather than just activity.
Key risks for investors
Lukasz de Pourbaix (15:41)
If we turn to risks—AI is clearly exciting, and anything linked to AI has generally performed well—but what are some of the key risks investors should be aware of?
Sue Lyn Stubbs (16:05)
The first is that AI amplifies existing risks. We’re already seeing this in the press.
There’s also the risk of “AI washing”—where companies overstate the value of what they’re doing.
One red flag is a disconnect between CEOs and boards. If management is very bullish but the board is more conservative, there may be misalignment on strategy and investment decisions.
We’re also focused on how companies measure return on AI investment. Some are now introducing more granular metrics—for example, moving from revenue per FTE to revenue plus capex per FTE, and tracking token usage across the organisation.
Another concern is that many boards cannot clearly articulate “no-go areas” for AI—whether that relates to bias, surveillance, or customer impact.
Finally, on workforce disruption, most boards suggested there would be little change over the next 12 to 24 months, which contrasts with some of the broader narratives.
Sam Heithersay (17:54)
Yes, I think there are some overinflated expectations around redundancies. Often those forecasts come from organisations with a vested interest in AI adoption.
One concept to keep in mind is the “J-curve” effect. Companies need to invest upfront—in data, systems, and training—before they realise efficiency gains.
So initially, AI can actually be disruptive and inefficient. Companies may even need to hire before they can reduce headcount.
At a broader level, we’re generally better at predicting jobs that will be lost than jobs that will be created. So there’s a tendency to overestimate negative effects.
From an investment perspective, this matters because large-scale job losses would have significant implications—but we think that adjustment will be slower and more gradual.
Sue Lyn Stubbs (19:46)
On that J-curve point, we spoke to the chair of a global financial platform who explained that they’re currently running dual systems—one traditional and one AI-enabled—because some customers don’t want to use AI.
As a result, costs have actually increased in the short term, and the expected efficiencies haven’t yet materialised.
So while there are productivity gains in parts of the business, overall returns are still unclear. It’s likely to remain uneven before benefits become more visible.
Sam Heithersay (20:40)
That’s an important point, and something investors are asking about now.
More advanced companies are starting to talk about balancing token costs with efficiency gains, including potential reductions in headcount.
When we conducted this work last year, the conversation was much more focused on upfront investment and increased tech spending. Metrics around token usage and efficiency trade-offs have only started to emerge more recently.
We expect those to become clearer, particularly among companies that are more mature in their AI capabilities.
Final thoughts for investors
Lukasz de Pourbaix (21:26)
As we wrap up, for investors trying to understand how to approach this theme, what would be your key guidance?
Sam Heithersay (21:51)
For us, it’s critical to stay very close to companies and management teams.
Ultimately, the long-term winners will be those that can adapt and pivot. Many companies will present compelling narratives around AI—it’s often well articulated—but it can be difficult to assess what’s real.
Comparing management teams and boards directly helps us understand differences in approach and capability.
Right now, it’s not obvious who the winners will be. That’s why we need to track this over time, continue engaging with companies, and refine our understanding.
Sue Lyn Stubbs (23:00)
I agree. The data simply isn’t available yet—we can’t rely on external proxies or standard datasets.
We need direct engagement with boards and companies to understand what they are actually doing, how they’re thinking about risk, and how they’re implementing AI.
That means looking beyond headlines and presentations, and really interrogating what these initiatives mean in practice.
This will be a key area of focus for us going forward—tracking progress over time and continuing that engagement.
Lukasz de Pourbaix (23:54)
Well, Sue Lyn and Sam, thank you very much for your time. That was a fascinating discussion and a very interesting perspective on the AI theme.
To me, it really highlights the importance of bottom-up research and being close to companies. The access we have is a real advantage in gaining these insights.
For listeners who are interested, as mentioned earlier, there is a paper on the Fidelity website, fidelity.com.au, called Governing in the Age of AI. Please refer to that if you’d like to learn more.
Thanks again.
Sue Lyn Stubbs (24:29)
Thanks.
