AI and the markets: boom, bubble, or breakout?

This transcript has been edited for clarity and readability.

In this webinar

• Welcome and panel introductions (00:03)
• AI adoption and where value is being created (02:41)
• Lessons from the dot-com boom (04:18)
• Bubble risk, market concentration and major AI IPOs (07:12)
• Opportunities across the AI value chain (14:10)
• AI investment opportunities around the world (18:21)
• Audience questions: long-term investing and business adoption (21:40)
• Open-source AI, infrastructure constraints and market risks (27:25)
• Overlooked opportunities, societal impacts and regulation (33:42)

Welcome and panel introductions (00:03)

Victoria Scholar

Hello everyone, and a very warm welcome. Thank you for taking the time out of your busy day to join us for this lunchtime webinar.

My name is Victoria Scholar, head of investment at interactive investor, and I’ll be your host today.

Our topic is ‘AI and the markets: boom, bubble or breakout?’ There is a lot to discuss. AI has become one of the dominant forces driving markets in recent years, particularly since the launch of ChatGPT at the end of 2022.

The AI value chain includes companies developing foundation models, semiconductors and cloud infrastructure, as well as businesses in traditional sectors such as pharmaceuticals, banking and energy that are harnessing the AI-driven economy.

Today, we want to explore how to invest successfully in an AI-driven world, including some of the opportunities, risks, winners and losers.

Before we get started, a little housekeeping. We aim to run for around 40 minutes. We’ll introduce our panel, have a discussion around the table and then answer questions from the audience.

If you would like to ask a question, head to slido.com and use the code 7469395. You can also submit a question through the YouTube comments. If you see a question on Slido that you would also like answered, give it a thumbs up to move it up the popularity leaderboard.

Finally, this webinar is for educational purposes only and does not constitute financial advice.

Our panellists are Chris Ford, portfolio manager and head of growth equities at Landseer Asset Management, and Jamie Mills O’Brien, investment director in developed market equities at Aberdeen.

AI adoption and where value is being created (02:41)

Victoria Scholar

Jamie, when most people think about AI, they probably think about ChatGPT, Anthropic or Nvidia, but AI is much broader than that. How do you think about the theme?

Jamie Mills O’Brien

Firstly, what we’ve witnessed is huge. AI is affecting the macroeconomy and markets, and the scale of investment is massive.

Over the past three and a half years, we’ve seen two types of AI emerge. The first is consumer AI, where adoption has probably been faster than expected. The best-known platforms, including Claude and ChatGPT, have around one billion monthly active users between them.

The second is enterprise AI, where the speed of adoption has probably disappointed expectations.

When we look at previous technology cycles, value has generally started in the hardware stack before moving into infrastructure and application software. We expected something similar to happen here.

What has surprised us about this generational technology shift is how long value has continued to accrue in hardware and semiconductors. That part of the market has captured most of the economics so far. The model developers are also making strong margins at the moment, although we can debate whether they will retain them.

The big question is whether the current supply-and-demand imbalance will correct and allow value to move into other parts of the technology sector.

Lessons from the dot-com boom (04:18)

Victoria Scholar

Chris, how does this transformational shift compare with previous major changes, particularly the internet and the dot-com era? What are the similarities and differences?

Chris Ford

There are notable similarities, but also important differences that sometimes get lost.

We’ve seen semiconductors and the associated hardware supply chain benefit significantly over the past three or four years. If you look back at the 1990s internet bubble, however, there was a five-year period during which stock-market performance became increasingly concentrated in telecommunications equipment and optical-component companies. That provides a useful roadmap for what is happening this time.

It shouldn’t necessarily surprise us that, three or four years into this cycle, value is still emerging and stocks are still being rewarded in that part of the market.

In the 1820s, you couldn’t run trains until tracks had been laid. In the 1990s, you couldn’t develop internet-based business models until the internet infrastructure existed. In the same way, we can’t build an artificially intelligent world until we’ve built the necessary infrastructure.

One of the key lessons from the internet bubble is that, although a huge amount of value was created and subsequently disappeared when the bubble burst, the greatest creation of value happened afterwards.

Companies such as Amazon and Google couldn’t have existed without the internet infrastructure built during the 1990s. You don’t need a search engine without an internet, and you can’t have ecommerce without one.

We think the same phenomenon will occur during the AI cycle. Over the next decade, once sufficient infrastructure is in place, the really exciting developments can begin.

Jamie Mills O’Brien

One important lesson is that the builders of infrastructure aren’t always the ones that ultimately create the most value from it.

The companies that laid fibre during the internet buildout generally didn’t create the greatest value. Software companies capitalised on the infrastructure instead, allowing them to grow without investing much additional capital and to earn very high margins.

Microsoft didn’t own the internet, but it was still a huge creator of value. We don’t yet know exactly how that analogy will apply to AI, but it’s an important lesson.

Bubble risk, market concentration and major AI IPOs (07:12)

Victoria Scholar

If we’re concerned about repeating the mistakes of the past, could we be heading towards a major bubble that bursts before AI transforms the economy in ways we can’t yet imagine?

Jamie Mills O’Brien

That’s a central question. One positive difference is that there is healthy scepticism and we’re actively discussing the possibility of a bubble, which perhaps wasn’t happening to the same extent during the dot-com era.

In public markets, we may be seeing a bubble in earnings rather than valuations. Private markets may have more of a valuation bubble, but the rise in public markets has been largely driven by earnings and fundamentals.

Over the next few years, different parts of the AI market are likely to perform differently. Some parts of the market may have overearned during this period, although that certainly doesn’t apply everywhere.

Another lesson is that the current cycle has initially been driven by semiconductors, which are cyclical. The industry has a long history of underbuilding and overbuilding, and it is unlikely that this pattern will disappear. We’re currently seeing explosive demand and remain in a supply-constrained world, but that will eventually change.

Victoria Scholar

Is that why only a small number of stocks are outperforming, despite AI being described as a global transformational force?

Chris Ford

The market has become narrower as developments in AI have unfolded over the past three years. We are now beginning to see early signs of how that may broaden.

Some of that comes from IPO activity, which is bringing new opportunities to the market at an extraordinary scale. SpaceX floated a couple of weeks ago, and regardless of what people think about the company, we’ve never seen an IPO quite like it.

We’re also seeing geographic opportunities emerge outside North America. For much of the past 30 years, investors may have felt they didn’t need to look much beyond the west coast of the United States to find innovative companies capable of shaping the next few decades.

That is no longer the case. There are excellent opportunities in Greater China and elsewhere. As the benefits of AI infrastructure become more widely available, I think a broader market will be characteristic of the next decade.

Victoria Scholar

SpaceX isn’t the only major IPO expected this year. OpenAI and Anthropic could also float. Jamie, how significant could these listings be for markets?

Jamie Mills O’Brien

We’ve already had a fair amount of equity-market activity, and SpaceX gave us an early taste.

These potential listings will be important barometers because there is a healthy debate about the strength of the frontier-model ecosystem. They will give public-market investors greater visibility of the financial health of these businesses, particularly as we assess the impact of open-source models.

These are capital-intensive, highly competitive industries. Anthropic may have a significant cost advantage over ChatGPT. It has also moved successfully into products such as Claude Code, and we’ve been surprised by how sticky these products have been, including among non-technical users.

That has implications for software companies, frontier-model developers and the overall health of the AI investment theme.

Chris Ford

Investors will need to consider the competitive advantages these companies do or don’t possess.

We largely believe that large language models will become commoditised over time. In that environment, how does a provider of frontier models compete? It will depend on other sources of differentiation: the quality of its software development kits, the strength of its customer list and the breadth of its plans.

Microsoft has arguably reinvented its AI strategy for the second or third time in a decade. It doesn’t own a frontier large language model, but it increasingly appears to understand its potential role as a gateway through which enterprises and their data can access whichever model they prefer.

Victoria Scholar

Jamie, why has the performance of the Magnificent Seven been so mixed this year? Mega-cap technology isn’t the same thing as AI, but there is a lot of overlap. Why, for example, is Microsoft down 20% since the start of the year?

Jamie Mills O’Brien

These companies used to be what we call capital-light compounders. They could grow without investing much capital, which was a wonderful business model during a decade of falling interest rates and disinflation.

That has changed. Their business models, including Microsoft’s, have become much more capital intensive. There has been a huge transfer of free cash flow from software and the Magnificent Seven towards technology hardware and semiconductors.

Two things have challenged Microsoft and others. The first is capital intensity. Will the enormous sums being invested now generate returns on capital as high as those produced by their core businesses? That uncertainty widens the range of possible outcomes.

The second is competition. Companies such as OpenAI and Anthropic are challenging parts of the existing business model.

One lesson from the past is that the eventual winners aren’t always those that own the infrastructure. They are often the businesses that control the data and customer experience. The market is still deciding whether Microsoft and others can control that customer experience in an AI world.

AI has been a fundamentals-driven story. Earnings and margins have driven much of the performance, rather than valuations alone. The reverse is also true: the market has reacted negatively to significant downgrades in free cash flow as Microsoft and others have increased capital expenditure.

Opportunities across the AI value chain (14:10)

Victoria Scholar

Which parts of the AI value chain are you most excited about?

Jamie Mills O’Brien

Semiconductor capital equipment is one area we find particularly interesting. That includes companies such as ASML, Applied Materials, Lam Research and Besi.

This was once an industry growing at around 7% to 9% a year; it is now growing at more than 20%. These businesses occupy important bottlenecks and are exposed to memory, logic and advanced packaging—the key technologies enabling AI.

They are high-quality businesses capturing a significant proportion of the economics. In other parts of the AI value chain, business models can be more speculative and the underlying economics and returns on invested capital less clear.

We also see many opportunities outside the US. A lot of the beneficiaries of rising capital expenditure are in North Asian markets, particularly Taiwan and South Korea. We believe there are strong technology companies whose prices don’t fully reflect their technology or position in the value chain.

If open-source models drive down costs and make the market more commoditised, volumes and returns could increase across the ecosystem. In that scenario, infrastructure providers and hyperscalers could benefit because they are exposed to the volume of activity rather than frontier-model margins. Some are trading on earnings valuations not far above the wider market.

Victoria Scholar

Chris, are there any lesser-known AI-related companies in your fund that you would highlight?

Chris Ford

We run what is effectively a long-only global equity fund. Although it is an AI fund, we invest in companies around the world whose economic futures depend on how they engage with AI. We find those businesses across the economy.

One example is NAURA, a Chinese semiconductor capital-equipment company. I agree that semiconductor capital equipment is interesting, but there is an important nuance: China is beginning to build its own domestic manufacturing ecosystem in this area. We think that will become increasingly significant.

Another example, from a seemingly distant part of the economy, is Veolia. It is a long-established French water, waste and energy-management company that has performed well over the past 12 months.

A change in management brought a new perspective on Veolia as a creator of data. Its assets include the pumps, pipes and waste-collection vehicles that people normally associate with the company, but it also holds an enormous repository of data.

Companies in traditional industries that can harness their data, extract insights and strengthen their competitive position have a substantial opportunity to create value that investment markets may overlook.

These companies may not produce the eye-catching daily movements seen in more volatile sectors such as semiconductors. Instead, they can produce steady outperformance quarter after quarter, which can be an attractive way to deliver investment returns on a volatility-adjusted basis.

AI investment opportunities around the world (18:21)

Victoria Scholar

Are particular parts of the world adopting AI more quickly, and is that influencing their equity markets?

Chris Ford

AI is everywhere. No part of the world or sector will be untouched by it, although there is a huge disparity between winners and losers.

We don’t believe traditional economic sectors are the best way to identify those winners. You need to examine individual business models and understand how companies are using data science and AI to strengthen their operations and competitive advantages.

Culture also matters. A company must be willing and able to change if it is to engage effectively with AI.

Not all companies or countries are equal in this respect. The rate of innovation among companies in Greater China, for example, is much faster than in North America and particularly Western Europe. That partly reflects the regulatory framework and policy direction.

The Chinese Communist Party recently required companies to have a coherent AI strategy by the end of this decade, explain it and demonstrate its effect on their business models. We’re still waiting for something similar from the European Union, and may be waiting for some time.

Victoria Scholar

What about the UK? Are there exciting AI opportunities closer to home?

Chris Ford

The UK punches above its weight, particularly in foundational research and technology. The problem for AI in the UK is the same one the country faces more broadly: how to turn a strong start-up culture into a successful scale-up culture. We are not very good at scaling businesses.

Arm, based in Cambridge, has been one of the great UK innovation successes of the past 35 years. We believe it could benefit significantly as AI inference increases the number of central processing units required across the economy.

The UK also has an interesting opportunity in quantum computing, although the market is still at a very early stage and most listed quantum companies are currently in North America.

There are some interesting private companies in Cambridge that could reach public markets over the next few years. One example is Riverlane, a leader in quantum error correction.

The attraction of Riverlane is that investors don’t necessarily need to predict which underlying type of qubit will win. Whatever form the qubit takes, quantum computers will require error correction on top of it. That can reduce some of the underlying technology risk.

Audience questions: long-term investing and business adoption (21:40)

Victoria Scholar

We’ll now move to questions from the audience. Our most popular question is: “How should we think about AI as an investment? Is it a long-term holding or a short-term trade?”

Chris Ford

AI is a very long-duration theme, but the nature of that theme will change over time.

We launched our fund in 2017, five years before ChatGPT existed and before the transformer model had been invented. AI has evolved over at least 70 years, and the companies investors may want to own will continue to change.

When investing in global equity markets over the coming decades, it will be important to understand each company’s position in relation to AI. If a business doesn’t have a robust, defensible and far-sighted approach, its chances of succeeding in its industry will be greatly diminished.

Jamie Mills O’Brien

I agree. It comes down to the individual company and whether its business model and competitive moat allow it to capture the economic benefits.

The theme is already changing. Hyperscalers initially dominated; memory producers are now beginning to capture both investor attention and stronger fundamentals. Infrastructure assets are also increasingly viewed as crucial and difficult to replace.

AI is a long-duration theme, but it will also be cyclical. The initial phase has been driven by semiconductors, which are cyclical assets. That is one of the main risks on which we are increasingly focused.

Victoria Scholar

Paulo asks: “Looking beyond semiconductors, chips and memory, which market sectors stand to benefit most from implementing AI and generate the returns needed to justify this huge infrastructure investment?”

Jamie Mills O’Brien

I agree with Chris that sectors may not be the best way to think about it because AI could affect them all.

Much of the difficult groundwork for enterprise AI still needs to happen: organising databases, working with small language models, and deciding where optimisation takes place and who controls it.

That helps explain the gap between optimism among senior executives and frontline workers, where many of the expected use cases haven’t yet materialised.

If those elements come together and enterprise use cases broaden, AI will become a genuinely cross-sector force.

During earnings seasons, people sometimes assume that falling cost ratios at banks must be the result of AI, but the evidence is not yet conclusive. We expect more companies across more sectors to identify real productivity gains.

Financial services have shown some early promise, while customer service and coding are among the areas where a clear product-market fit has emerged. Ultimately, AI needs to produce either savings or additional revenue to generate a return on investment. We haven’t yet seen that happen at the required scale.

Chris Ford

There are some strong early examples of returns, even from before transformer models and large language models became dominant.

A 2019 report suggested that European insurers could potentially improve margins by around two percentage points if they deployed the AI technology available at the time thoroughly across their organisations.

Insurance isn’t an industry renowned for moving quickly. One challenge is that organisations need to understand their own business processes before they can apply AI effectively.

Many processes are poorly documented or understood, are decades old, or were designed by people who have since left the business. If a company doesn’t understand what it is doing, it is difficult to apply AI to improve it.

Jamie Mills O’Brien

Another question is which profit pool AI will disrupt. Google disrupted print advertising and Amazon disrupted traditional retail. AI is expected to disrupt labour, but we haven’t yet seen that happen at scale.

AI may also create entirely new markets. However, as annual capital expenditure rises from around $750 billion to $850 billion, the need to demonstrate use cases and productivity savings becomes increasingly demanding. That helps explain some of the volatility we’re seeing in markets.

Open-source AI, infrastructure constraints and market risks (27:25)

Victoria Scholar

Chris, how could the proliferation of inexpensive, high-performing Chinese open-source AI models disrupt US big technology companies?

Chris Ford

It depends partly on the extent to which the US can engage with them, which makes regulation important.

Investment markets have generally overreacted to innovation from China. DeepSeek was the most obvious recent example, although investors are becoming more knowledgeable about the issue.

Many Chinese models have been trained partly using foundational work carried out in North America. Nevertheless, the competitive lines are beginning to be drawn.

A large number of US chief executives recently signed an open letter calling for large language models to be as open source as reasonably possible.

The signatories included people you might expect, such as Nvidia chief executive Jensen Huang, but also frontier-model developers such as OpenAI and Anthropic and leaders from the previous software cycle, including Microsoft. Few companies know more about operating a closed-source software monopoly than Microsoft.

Jamie Mills O’Brien

It will be interesting to see how the US midterm elections and potential AI-safety legislation affect regulation and the protection of the domestic AI industry.

Jensen Huang’s support for open source may help explain the possible direction of the market. Frontier models could perform the hardest tasks, where errors are unacceptable, while cheaper open-source models handle simpler tasks that don’t require the very best performance.

Jevons paradox suggests that when costs fall, usage can increase significantly. Open source could therefore support much higher volumes, clearer returns on investment and more use cases.

Nvidia supplies much of the infrastructure behind all of this. It also doesn’t want to depend on one or two customers; it would rather sell to many. A broader open-source market could suit Nvidia.

Victoria Scholar

Another attendee asks: “How much will the availability of power infrastructure—including grid transformers and backup equipment—hold back the huge AI rollout? Should investors focus on the companies supplying the tools rather than the finished product?”

Chris Ford

Power availability is already a constraint, and countries are not equally placed. The UK is arguably in a relatively poor position because of its power infrastructure, while other countries are better prepared.

We would already have more data-centre capacity if electrical generation were more readily available and developed economies had invested more in their grids over recent decades.

However, it is a false choice to say investors must choose either infrastructure or finished products. Both can offer opportunities, depending on the company.

One feature of the power-infrastructure market is that companies such as GE Vernova are close to the limit of their manufacturing capacity. If you want a gas turbine, you may not be able to obtain one before 2031.

Investors still need to consider when that strong demand will appear in a company’s financial results. If fundamentals continue to drive the market, those fundamentals must move in the right direction for the shares to perform.

The broader question is where the next infrastructure bottleneck will arise. At different times, the constraint has been memory, copper, generating capacity, wiring or fibre optics.

It isn’t enough to identify the current bottleneck, because the rest of the market can see it too. Investors taking a shorter-term, cyclical approach need to anticipate what comes next.

Victoria Scholar

How real is the risk of a market crash?

Jamie Mills O’Brien

There are several ways to approach that. Corporate fundamentals, particularly in the US, remain robust. The recent wobble appears to have been more technical than fundamental, and the current earnings season has broadly confirmed that the AI sector remains healthy.

One risk is that some of the bottlenecks Chris mentioned—including permitting and power equipment—eventually disappear.

TSMC is currently the main semiconductor bottleneck. There are no signs that it is behaving irrationally by building excessive capacity, and neither are alternative suppliers such as Samsung and Intel. That could change, at which point our view would also change.

Because the theme has been driven by fundamentals, earnings have risen sharply in many places. Valuations for some high-quality companies therefore don’t look completely excessive, despite the noise surrounding AI. Some parts of the market remain reasonably valued.

Overlooked opportunities, societal impacts and regulation (33:42)

Victoria Scholar

An attendee asks: “Are there any overlooked opportunities in AI that could become successful in years to come? Are there areas with potential that aren’t receiving as much attention?”

Chris Ford

The answer has to be yes.

Market performance has been relatively narrow. Until very recently, almost anything outside memory—or diversification away from memory—has been detrimental to performance.

You could therefore argue that many businesses, including some within technology, aren’t receiving proper recognition.

Some companies have also been caught in the sell-off in software over the past year, even though they shouldn’t be written off. The idea that software as a whole has been defeated is extremely premature, and that creates opportunities.

Software was treated as different, better, more robust and longer duration for around 15 years. What we are now seeing may partly be the reversal of that misguided perspective.

AI will disrupt the whole economy, including software, but that doesn’t mean every software company will lose. There will be winners.

Jamie Mills O’Brien

I agree. The way the market has treated certain parts of the economy is creating opportunities.

Infrastructure software is one area we find interesting. Businesses such as Datadog, Snowflake and MongoDB could play important roles in the groundwork required for enterprise AI. They are strong companies, but their shares often fall whenever the wider software sector declines.

We also see opportunities in semiconductor capital equipment and interconnect technology, which links GPU systems and data centres. This part of the supply chain has been underinvested in and is becoming more important. China also plays a significant role.

Chris Ford

There are opportunities in utilities, insurance and even parts of consumer staples. Businesses in these traditional sectors could use AI to improve their models significantly.

If AI allowed a business with a 10% operating margin to increase that by two percentage points, it would be transformational.

Jamie Mills O’Brien

One of the most interesting developments for technology investors is the return of the old economy.

We are used to seeing capital-light technology companies perform well, but older, irreplaceable assets have suddenly become vital. Utilities and grid-infrastructure providers are good examples. Their assets have always been important, but the market is now recognising that more clearly. We think that could continue.

Victoria Scholar

Our final question is: “What major societal effects do you expect from emerging AI technologies, and how are markets pricing them?”

Jamie Mills O’Brien

We’re seeing unprecedented demand for capital across public and private markets.

Governments are investing in infrastructure and critical technology while increasingly restricting how those assets are shared with allies. At the same time, companies such as Google are raising debt and equity to fund huge capital-expenditure programmes.

That demand is increasing the cost of capital, limiting valuations and making earnings more important.

We haven’t yet seen the full effect of AI interacting with real-world applications, which we suspect will take longer to emerge.

Another consideration is employment. Much of the spending assumes that AI will displace jobs and change the labour force. At present, we think productivity improvements are more likely than widespread job losses.

Chris Ford

At a higher level, the regulatory framework within which AI will operate remains ill-defined and largely undecided.

Regulation reflects the philosophical traditions of the cultures creating it. In the West, we tend to believe personal data belongs to the individual and can be used only with permission. In China, data is more likely to be viewed as something that can be used for the greater good unless there is a reason to keep it private.

China’s approach reflects thousands of years of Confucian tradition. Understanding how these different philosophies shape regulation—and create tangible competitive differences between regions—will be essential.

Jamie Mills O’Brien

AI appears to have arrived at a time of deglobalisation and greater emphasis on domestic priorities.

During previous technology transitions, countries shared more with one another, particularly with allies. Now, governments increasingly want to protect technology and data.

How AI regulation develops against a backdrop of weaker supranational institutions, deglobalisation and changing supply chains will become central to how the technology evolves.

Closing remarks (40:01)

Victoria Scholar

Thank you both. We’ve tried to squeeze a great deal of information into 40 minutes, and our panellists have done an excellent job.

A huge thank you to Jamie Mills O’Brien, investment director at Aberdeen, and Chris Ford, portfolio manager at Landseer Asset Management.

We hope to see you again for another webinar soon. We’d love to receive your feedback on what you liked or disliked so we can continue tailoring these sessions to you. You’ll receive an email later today where you can share your thoughts.

If you don’t already subscribe to our YouTube channel, please do. Thank you, and we’ll see you again soon.

The panellists described AI as a long-term investment theme that will change over time. Different types of companies may benefit as technology advances, while sectors such as semiconductors could experience shorter-term cycles of rising and falling demand.

They suggested assessing individual companies based on their business models, competitive advantages, and ability to capture the economic benefits of AI.

AI is expected to affect almost every sector, making individual companies more important than broad sector classifications.

Early applications include coding, customer service and financial services. The panellists also identified potential opportunities in insurance, utilities and consumer goods. However, businesses will need to demonstrate that AI can generate additional revenue, reduce costs or improve productivity.

Lower-cost, open-source Chinese models could increase competition and make AI more widely available. Frontier models may continue to handle complex tasks where accuracy is critical, while cheaper models perform more routine work.

Falling costs could also lead to greater adoption and more demand for the infrastructure supporting AI. This means some US technology companies could face increased competition while others benefit from higher usage.

Power generation and grid capacity are already restricting the construction of data centres in some markets. Other bottlenecks include transformers, generating equipment, memory, copper and fibre-optic infrastructure.

The panellists said this could create opportunities for infrastructure providers, but investors still need to consider whether strong demand will translate into higher earnings, and when.

The panellists said corporate earnings and demand for AI infrastructure remained relatively strong at the time of the webinar. Some company valuations therefore appeared to be supported by earnings growth rather than speculation alone.

However, risks remain. Semiconductor companies are cyclical, businesses are committing vast sums to AI infrastructure, and investors will increasingly expect evidence of productivity gains and returns on that spending.

The panellists highlighted potential opportunities beyond the most prominent chip and technology companies. These included enterprise software, semiconductor manufacturing equipment, data-centre interconnect technology, utilities and businesses in traditional industries that can use their existing data more effectively.

They argued that AI will disrupt software and other sectors, but this does not mean every company will lose. The challenge is identifying businesses with valuable data, strong competitive positions and credible plans for using AI.