In February last year we released a new whitepaper titled ‘The Amazing hyperscalers’, which foreshadowed that artificial intelligence (AI) was the market’s most important investment theme.
Since then, ChatGPT exploded onto the scene and has been wowing users with its ability to almost instantly create human-like text. Other AI products are now producing video, transcribing, writing code and designing to remarkable standards.
The hype around ChatGPT and AI has finally focused investors on how to best profit from this trend. But it has also raised serious questions around the broader economic and social impacts of AI.
In this Q&A, we update our thinking on AI by addressing 10 vital questions ranging from the likelihood of job losses, to the prospect of another ‘dot-com’ style bubbles, and the stock market winners and losers.
1. In your report released in early 2022, ‘The Amazing hyperscalers’, you defined the AI revolution as ‘epoch defining’ and today’s most important investment theme. Where does AI fit in the pantheon of technological development?
We don’t really know for sure as yet. But the extraordinary advances in AI over the last 10 years – and more importantly, the rate of change of these advances – suggest that AI will likely be revolutionary, not evolutionary.
Tech strategist, Ben Thompson, recently put it like this:
- The PC enabled information to be duplicated at zero marginal cost. (Marginal cost is the increased costs of producing additional units. Zero marginal cost means there is no additional cost to producing more units.)
- The internet enabled information to be distributed at zero marginal cost.
- AI enables information to be generated at zero marginal cost.
This is a good way to think about it – especially as we all experiment with ChatGPT and text-to-image generators and the like.
But the real value of AI over time will likely stem from its general problem-solving capabilities. We’re not there yet, but AI will likely represent something like a ‘skeleton key’ that will unlock the solutions to the world’s scientific, medical, and engineering problems.
This is why some experts much smarter than myself believe that ‘solving’ Artificial General Intelligence (AGI) – that is, building a model, or collection of models, that are general problem solvers – will be the last problem that humans ever solve.
2. What broader economic benefits do you expect from AI? Will it enhance productivity and increase economic growth rates?
I think it will unambiguously increase the productivity of humans. Already, tools around content generation – whether in text, images, or code, such as transcription and translation, just to name a few obvious ones – are drastically reducing the time and cost associated with many tasks.
Some of the first academic research found that customer-support staff increased their productivity by 14% when using AI-based conversational assistants. Though, interestingly, the productivity gains were all captured by novice and low-level employees – not the more highly-skilled workers.
As for economic growth rates, I’m not sure we can answer this today. I really don’t think the answer is obvious. On the one hand, greater productivity is helpful to growth, all else equal. But all else probably won’t be equal over time.
For example, in the hypothetical extreme (just as a thought experiment), if labor is essentially no longer needed for most professional tasks, then we end up with most of us being unemployed with no source of income. And the bulk of the economic spoils of this new economy will accrue to a small number of companies which own the AI models.
This simple thought experiment suggests that, at a minimum, economic inequality will likely increase. And we know more extreme inequality is actually a headwind to consumption growth (and, therefore aggregate economic growth) because, while most people spend the majority of their earnings on goods and services which keeps the economy growing, the wealthy few simply can’t spend enough of their wealth – especially as it expands.
I think this is why many experts who have been thinking about this future scenario for a long time believe that societies will need to adopt some kind of Universal Basic Income (UBI) over time. A UBI is a guaranteed minimum level of monthly income for all citizens and is provided without condition and irrespective of any other form of income. Such an idea may sound really strange at first, but empirical studies are showing lots of counterintuitive benefits around greater risk-taking and entrepreneurship.
3. Does AI have the potential to trigger another dot-com style stock market boom?
Sure, there’s potential. But it will probably look and feel a bit different.
So, today, for example, we see some AI-related stocks that look overvalued due to the hype. We also see some AI leaders, including Microsoft, Amazon and Meta, that look substantially undervalued – and, of course, these are the ones we try to own.
And we see lots of business – in nearly every industry – that are going to be impacted by AI in a meaningful way. Some positively and some negatively.
So our sense is that this transformation is going to split winners and losers in a much more definitive way than the dot-com boom did 25 years ago.
4. The ‘Hyperscalers’ report flagged that the big winners from AI would be the ‘hyperscalers’ Amazon, Alphabet and Microsoft. Their stocks have rallied as investors caught up to the thesis and have realised the massive potential of AI. Has that rally got further to go?
For Microsoft and Amazon, the answer is yes – the rally has a long, long way to go, in our view.
For Alphabet, the answer is also yes – but with the caveat that there is near-term risk in the profitability of its core search business, so there could be some bumps along the way.
5. Are there any smaller AI stocks that are attractive?
The answer is unquestionably yes, though I don’t know where they all are. (Please email us the answers).
One neat example of a high-probability AI winner that is relatively smaller in size (i.e. US$12 billion in value) is Unity Software. They own the video gaming engine upon which more than 70% of the games built on third-party tools outside of Asia are developed and run.
They are deploying AI in lots of ways: generative tools to increase the productivity of game development; increasing the realism/authenticity of in-game experiences; and predictive tools that improve in-game advertising targeting and measurement.
Their gaming engine is also being used increasingly by other industries – from commercial real estate, to manufacturing, to energy – to model ‘digital twin’ simulations of buildings, or plants. A digital twin is simply a virtual digital model designed to accurately reflect a physical object. Unity is deploying AI models here as well to make real-time predictions that can aid property managers’, or plant managers’ decision-making on the ground.
And just like Microsoft – though at a smaller scale – Unity is set to benefit both from
- the ‘distribution’ of AI to its customer base in the form of high-profit-margin incremental revenues, and
- incremental revenues tied to the ‘compute consumption’ where its customers inference (i.e. send query requests to) the AI models stored on their servers over and over again.
6. In the last Monocle, you said that not only winners from the AI revolution are emerging, but also losers. Who are the major losers from AI and what should investors avoid?
One group of losers will include those businesses that were built on charging a lot of money for services that can now be performed by AI models at negligible costs.
Last month, for example, a business in the US called Chegg, an online educational platform that assists students with their homework, told investors they now believed ChatGPT was having “an impact” on new customer growth – and not in a positive way. The stock basically halved in a day. There will probably be a lot more of these situations over time.
Losers will also likely include businesses that refuse, or are slow, to adopt these new tools and evolve their business processes.
For example, one way to think about it is something like this: if my competitors can all deliver the same product at a greatly reduced cost thanks to AI-based applications, then I had better reduce my costs too, or else the inevitable price deflation that’s coming will wipe out my profits.
7. Do you expect AI to make many jobs redundant?
My current thinking is that, over the medium term, AI will likely make some jobs redundant (obvious examples include transcribing jobs, or document translating jobs), and it will contribute to some significant wage deflation in other jobs – particularly white collar mid-level jobs.
This view relates back to the early evidence we are seeing around productivity. In some white collar jobs, a novice employee using AI tools can perform around the level of a more senior employee. This points to downside risk in the wages of that type of senior employee as the pool of available labor for that role effectively increases significantly.
I also think that, as a general rule, team sizes inside corporates will generally shrink. And this is because, as humans become more productive, armed with their various AI tools, the ‘bottleneck’ to progress will increasingly become the politics between the humans. So relatively smaller, collegial, multidisciplinary teams that can make decisions quickly will probably become the template for success over time.
So just like for businesses, there will be winners and losers in labor over the medium term. And to where this all leads in the long run, who knows?
8. Montaka’s latest report, ‘Why reports of tech’s death are greatly exaggerated’, argued it was unlikely that regulators would crack down on tech profits. How do you see governments and regulators dealing with AI?
I think governments are starting to become more aware of the opportunities that AI can offer, as well as the risks or dangers of the technology.
And this is really encouraging – after all, it’s not easy to stay up to date in a field that is very technical to understand, is changing so significantly, so quickly, and has potentially massive implications across a wide range of societal dimensions.
I think Australia has an enormous opportunity here, enabled by our healthy democracy and functional working relationship between the two major parties. Australia has a really good shot at capturing the best of what AI has to offer, while mitigating the dangers, and supporting its society through the change.
I currently live in the United States and I’m far less optimistic that the two major political parties could agree on how to deal with the very large societal changes that may well be on the horizon.
9. Can you see AI solving some of the world’s big problems like inequality and climate change?
I see AI making inequality more extreme, unfortunately. There will likely be a small number of big winners, and large number of losers, along the dimensions of business, labor, and probably nations as well.
On climate change, I think yes, unquestionably, AI will directly contribute to helpful solutions. It already has – in electric vehicles and smart energy-efficient buildings, for example – and this will only continue.
As for ‘solving’ climate change, it seems to me that it’s a race against time. Fortunately, advances in AI are increasing in such an extraordinary, non-linear way, that I’m hopeful on this dimension.
10. How can investors use AI themselves to invest in the stock market?
Well, I wouldn’t recommend investors outsource their investment decisions to ChatGPT, if that’s what you mean.
Here’s how I would frame it for investors. Let’s say we broadly disaggregate the investment process between:
- ‘production’ – which is all the required reading, writing, and analysing; and
- ‘insight crystallisation’ – which is all about thinking deeply and creatively about implications, assessing probabilities of possible scenarios, and ultimately arriving at high-probability perceptions that are different to those of the market’s and represent investment opportunities.
AI is going to be of great assistance to investors for tasks in the production category. It will allow investors to go much wider and deeper in a much shorter amount of time.
And my recommendation would be to go wider and deeper for sure, but also reinvest some of those time savings in more insight crystallisation. Ultimately, I believe this is where the value-add will remain for human investors for the foreseeable future.
To request a copy of our latest whitepaper on Tech & know the top 5 stocks poised to lead the tech recovery, please share your details with us:
Montaka is invested in Microsoft, Amazon, Alphabet & Unity.