This market commentary is drawn from Brad Slingerlend of NZS Capital’s newsletter published on 31 May 2026, Stuff I Thought About Last Week (“SITALWeek”) #463
As we write this note at the end of May 2026, there are 14 companies with a market cap in excess of US$1T (NVIDIA, Alphabet, Apple, Microsoft, Amazon, Saudi Aramco, Meta Platforms, TSMC, Berkshire Hathaway, Broadcom, Tesla, SK Hynix, Samsung Electronics, and Micron). You might notice a pattern on that list: 11 of the 14 are in tech and telecom, and if we add Tesla (which is technically a consumer discretionary), that makes 12. Three recent entrants – Hynix, Samsung, and Micron – rode into the trillion-dollar club on the back of strong DRAM demand for AI computing. And, we are potentially poised to see a few new corporate members with the upcoming SpaceX IPO and market rumours of Anthropic and OpenAI IPOs perhaps later this year.
Looking more closely at market dynamics, this group of 14 companies is currently valued around US$30T, which is around 40% of the total US stock market cap and 20% of global stock markets. Notably, in the next tranche (US$500B-$1T) behind these giants there are only ten companies (Eli Lilly, Walmart, JPMorganChase, AMD, ASML, Visa, ExxonMobil, Intel, Oracle, and J&J), and their total value is around US$5.5T, which is roughly equal to the market value of the current largest company, NVIDIA.
As students of complex adaptive systems, we are no strangers to power laws, and what we are seeing in today’s market is the result of numerous power-law winners creating and consolidating value – in terms of both dollar value in the stock market and real value for their customers. Like we’ve experienced with prior technological advances (PCs, the Internet, ecommerce, streaming video, smart phones, cloud computing, etc.), AI is rewriting the operating system of entire segments of the economy. This new fabric will transform every industry in unpredictable ways and is powered by a handful (now several handfuls) of mission-critical companies. We have been writing about the underlying dynamics of power-law winners in complex adaptive systems for decades, and here we will summarise the key elements.
In order to determine which platforms and suppliers will emerge as power-law winners in the AI Age, we think the following characteristics provide the most information:
- creation of increasing returns and network effects for users (i.e., the more a product is used, the better off all its users are);
- vertical integration, which tends to drive down costs and thus lure users, developers, and creators;
- adaptability – in times of disruption and early product development, the ability to change course is critical to success;
- access to capital – self-funding platforms and those that need less capital per dollar of revenue generated will have an advantage (which feeds their ability to adapt);
- and, most importantly, the degree of non-zero sumness (NZS) – i.e., the creation of more value than you take; high NZS companies enable customers, entrepreneurs, suppliers, and the ecosystem at large, thus cementing themselves as indispensable platforms, increasing their overall total addressable market (TAM), and insulating themselves against potential disruptors.
Importantly, they accomplish all of this while also being the lowest cost provider/tariff extractor in their ecosystem.
Historically, we have found these five factors indispensable for identifying emerging power-law winners, and, more importantly, they have helped identify which companies have a good chance of holding/expanding their natural monopolies vs. those that are vulnerable to disruption. As many folks are contemplating the current trove of trillionaire companies and potential IPOs, I’d like to focus on a few of these topics as they relate to the AI industry:
Network Effects
First, let’s start with: Will there be network effects for AI?
One of the defining qualities of the digital era is that the largest, most profitable businesses are winner-takes-most monopolies. This type of business ecosystem has evolved largely because a single system/algorithm is sufficient to feed all customers, creating benefits of scale that, in turn, benefit all users – the classic virtuous circle of win-win (this is one of the reasons we look for the highest non-zero-sum [NZS] outcome when we analyse industries and companies). For example, a cloud software company runs one version of their platform for all customers. The more customers creating data on that cloud app, the smarter the app becomes and the better off all customers are. Google search, as complex as it is, is one giant algorithm for everyone and improves as everyone uses it. But, AI, particularly interacting with agents, feels more individually tailored. This point was made by Microsoft’s head of AI, Mustafa Suleyman, recently: “we’re making arbitrary choices about the feeling of these models...the differentiating factors are indexed on the preference of the individual”. There seems to be network effects in both training models and the basic hardware used to run them; but, with customised interactions, is my AI agent getting better for me because of conversations your individualised AI agent is having with you? Maybe, but it seems like the network effects are abstracted from the user interface (i.e., the conversation) down to lower levels of the technology stack. That doesn’t mean AI won’t be winner takes most, but the winner in this case might be the lowest cost provider of the highest output AI, not the AI agent interface itself.
Vertical Integration
If the benefits of scale and increasing returns are going to come lower down in the technology stack, i.e., hardware, we should next address vertical integration.
In any technology platform transition, such as the current pivot to commoditised intelligence, there are going to be layers to the tech stack where more money will be made at different points in time than others. What is generally unchanged, cycle to cycle, is that the value distribution across the stack will be bar belled, with the new intelligence cycle being no exception: most money is made at the bottom (semiconductors) and the top (applications). The overall tech stack for the modern cloud and consumer world is roughly the following (with some omissions for brevity):
- Applications (Google Search, Uber, Netflix, Microsoft Office, Instagram, SaaS, etc.)
- Operating systems (LLMs, open source, MSFT, iOS, Android)
- Databases
- End-point hardware (mobile phones, PCs, connected devices)
- Communication (wireless, broadband)
- Compute hardware (servers, storage, networking)
- Chips (GPUs, CPUs, memory; semi-cap equipment and chip design software; connectors)
There are certain points in a new technology cycle where you can make money in any segment of the stack above. However, given the complexity of investment timing and all of the moving parts, the special products and services that seem to harness network effects and/or power laws to amass the largest markets tend to be near the bottom or the top (there are some exceptional monopolies that occasionally find their way into the middle, but the mid-stack layers tend to be least valuable and most vulnerable to disruptive cycles).
If this analogy holds, the LLMs – i.e., the operating system of the next wave of compute – may be less valuable than both the applications built on top of them (follow the developers!) and the foundational chips on which they run...We tend to find that vertical integration is key to creating the runaway, power-law winners, and I think this trend will hold true for AI – perhaps even more so than for the prior cloud computing platform shift.
The other thing we see that creates power-law winners, beyond network effects, is vertical integration. This started with the IBM mainframe; the more vertically integrated you were, the better positioned you were to win. This is very much true in the cloud, where Amazon built their own server chips and designed their entire tech stack from scratch to attract developers. However, what is interesting now is that among the vertically integrated platforms, there is a lot of differentiation—there are many winners and losers, and the outcome isn't totally clear.
At the moment, Google is the most vertically integrated player in AI. Transformer language models originally came out of search autocomplete. If anyone remembers search before it started filling in the query for you, you had to type the entire thing; that was the beginning of transformer models. All these large language models (LLMs) originated from a single paper published by Google in 2017.
Google has been building its entire cloud infrastructure, including its sixth-generation chip called the TPU, which is optimised for the tokenisation of language, specifically because of search autocomplete. They are in such a unique position to benefit from this that even OpenAI has had to come to them to run ChatGPT loads because it is significantly cheaper to run them on Google than on Microsoft and other platforms.
Amazon vs. Google: If we contrast Google’s position with Amazon (AWS), Amazon was effectively the runaway leader of the cloud. They have over 40% market share and more than half of the profits in cloud software. However, in AI, they are lagging dramatically with effectively no share. They are getting a little work, but mostly they are just growing their traditional cloud software business.
One of the reasons for this is that they did not build a vertically integrated infrastructure for GPUs or inference workloads for AI, so they were unable to attract developers early on. In contrast, ChatGPT working with Microsoft Azure, and Google with Gemini and their TPUs, were able to do that.
Let’s look at OpenAI. Are they vertically integrated? No. They sit at the top of the stack, acting as the operating system, but they are also trying to be the application (ChatGPT).
Our view is that the actual large language model is going to be a commodity. There are multiple frontier models, and the training burden isn't so high that we won't have a lot of competition. Value will likely be created by the applications that emerge on top of the LLMs, which is why you see OpenAI trying to build the applications themselves.
However, that strategy has traditionally not worked. Usually, you need the application first, and then you back your way into vertical integration. OpenAI doesn't control any of the stack below them. They haven't designed their own chips or data centres. They rely on Microsoft, Oracle, Google, and CoreWeave. With at least four infrastructure partners to optimise their workloads for, I don’t know if they will ever be cost-effective. Right now, the more revenue they generate, the more money they lose, and they are looking at needing to raise hundreds of billions of dollars in the next few years.
Future Outlook: If I had to place a bet right now—based on everything we know about the evolution of biological systems, network effects, benefits of scale, and vertical integration—it would suggest that OpenAI could end up as simply an asterisk in what was a very interesting time period for a technology platform shift. Five or ten years from now, we may not be using ChatGPT. That is a view I hold loosely, as we may go in another direction.
I will note that there are two sides to vertical integration to be aware of. Sometimes less vertically integrated companies can move more quickly to take advantage of a disruption. For example, Anthropic appears to have designed its AI to run on TPUs from Google as well as Trainium at Amazon, and they recently leased an NVIDIA-based data centre from Grok/SpaceX. If a large efficiency gain were to take place in another hardware setup, Anthropic could likely port over with little delay, whereas a vertically integrated company like Google would be unable to easily pivot from its TPU stack, potentially placing it at a cost disadvantage. Vertical integration can cut both ways. In most industries, the risk of ossification from vertical integration outweighs the benefit; however, in early technology platform shifts, it tends to be a net benefit. In the case of AI, it’s best to keep a close eye to see if vertical integration will continue to determine the low-cost winner or become a vulnerability.
Non-Zero Sumness (NZS)
Let’s move on to non-zero sumness, or win-win, created by the different AI platforms. This measurement is more complex because AI is replacing human intelligence, and, in so doing, may displace human jobs. AI is more of a punctuated equilibrium rather than a stepwise iteration. Unlike dotcom, smartphones, or the cloud, AI can think for itself, which introduces an entirely new level of chaos. Rather than merely democratising information, AI is rendering intelligence into a commodity. There will be winners and losers, but the primary trajectory of AI isn’t to make human jobs easier, but to replace humans at the executive functioning level. In prior shifts, technology has created productivity and some job displacement; but, machines subsuming intelligence…is perhaps the most chaotic input into this evolving complex adaptive system. AI that eats jobs too quickly may self-destruct by eliminating too much of its consumer base, but AI that can help create jobs (enabling creative new startups and scientific discovery) could blossom. Likewise, the winning AI platforms will be the ones that allow the most value to be created on top of themselves for the least amount of money, allowing them to accrue market share. We think vertical integration and cost efficiency will likely be the primary drivers for creating NZS versus network effects or other factors.
Analysis Notes
At NZS Capital, we evaluate the power-law indicators discussed herein as part of a larger survey of Quality, Growth, and Context to determine the potential long-term value of any business (see Chapter 3 of Complexity Investing). While we cannot comment publicly on any new or rumoured IPO specifically, this framework may assist investors in forming their own views. For example, with the current IPO of SpaceX, we would analyse their vertical integration, adaptability, non-zero sumness, network effects, etc. across Space, telco (the primary business today, soon to be supplanted by renting AI capacity to Anthropic), AI, and social media. When it comes to Context, we would consider Total Addressable Market (TAM) and valuation. Specifically, we would determine the necessary rate of return based on the company’s range of outcomes and look at where the TAM would need to be in a decade. From there, we could back into what percent of that TAM the company would need to achieve to earn the desired rate of return starting from the stock’s valuation today. If it’s reasonable, we would size that IPO position appropriately. If not, we are always happy to wait on the sidelines. At NZS Capital, we frame outcomes in terms of Resilience and Optionality (see Chapter 6), matching position size in the portfolio to the potential range of outcomes. It’s impossible to be precise on valuation in early-stage industries, so we prefer to focus on the bigger picture and the ability for a business to stack new S-curves on top of existing ones.
I would also note that several of the points above suggest a strategy of investing in the “picks and shovels” of AI – i.e., the chip and infrastructure layer – while taking a more cautious wait-and-see approach with respect to the value creation by the large language models. We also expect a Cambrian explosion of new products and services built on, and with the help of, AI. Given that everyone who has internet access also has access to AI (unlike prior tech waves that required infrastructure/hardware outlays or massive data relocation), this wave of entrepreneurship could be larger and faster than anything the economy has experienced.
IPO Market Impact
We get a lot of questions regarding the market dynamic effects of one or more large IPOs happening within a short timeframe. While the impact will be unpredictable, historically, large IPOs have not had a negative impact. According to Goldman Sachs’ April report titled: “Answers to 7 common client questions on potential impending mega IPOs”, the bank expects US$160B of gross IPO proceeds from 100 IPOs in 2026. The previous high was US$115B of proceeds in 2021. The GS data do not indicate meaningful selling pressure on other large stocks ahead of large IPOs; however, investors do tend to raise some amount of cash going into an IPO.
Typically, IPOs do not enter indices for many months after they start trading, and only then do they enter at their free float (i.e., the number of shares available, excluding shares held by insiders or shares subject to a lockup agreement). For SpaceX, the Nasdaq-100 Index will fast track inclusion 15 days after IPO at a rate of 3x the float. This could create elevated demand by index funds based on the Nasdaq-100. SpaceX also features a staggered lockup that could render more shares available to sell sooner than other typical IPOs. Other index makers are seeking feedback and may decide to fast track future large IPOs, and we may see more creative structures as new trillion-dollar companies are minted. While the mechanics and impacts are unpredictable, one thing is for sure: more trillion dollar companies are coming to the market in the near future.
The End, or Just the Beginning?
It’s difficult to examine the current state of the market without reference to the late 1990s, in particular the IPO frenzy. Measured against that scale, we are perhaps only on the cusp of AI mania. It’s possible that several large IPOs followed by many smaller IPOs could create a 1999-like moment at some point in the future. However, we are not yet at that juncture. While the market may seem concentrated, the nature of power-law winners in the digital age explains much of the current capital distribution. Combined with speculation around the ultimate size of the AI market and its impact on the economy, there still remains the potential for fat-tail outcomes to both the downside and upside. There are three ways we’d expect a downside tail event to occur:
- access to capital dries up, particularly as increased speculative lending is funding capex;
- models experience unexpected efficiency gains, tempering the pace of capex;
- regulatory pushback or other analog negative feedback loops (e.g., lack of available power, raw materials, etc.) impede progress.
As investors analyse the current megacaps and future IPOs, they would do well to monitor for these negative tail outcomes without losing sight of the potential magnitude of positive tail outcomes. All this is to say, the range of outcomes remains wide for the megacaps, but we believe the framework above is the proper guide for conceptualising both the upside and downside scenarios.
This information has been prepared for use only by wholesale clients and professional investors (as defined under the Corporations Act 2001 (Cth)).
Channel Investment Management Limited ABN 22 163 234 240 AFSL 439007 (‘CIML’) is the responsible entity for the NZS Global Growth Trust ARSN 691 841 335 (‘Fund’). Class B units in the Fund are distributed by Continental Funds Group Pty Ltd ACN 688 643 245 AR No. 1317974 (CFG), which is an authorised representative of CIML. Neither CIML, NZS, CFG nor their affiliates, officers, or employees make any representations or warranties, express or implied as to the accuracy, reliability or completeness of the information contained in this document and nothing contained in this document is or shall be relied upon as a promise or representation, whether as to the past or the future.
Past performance is not indicative of future performance. This information is given in summary form and does not purport to be complete. Information in this document should not be considered advice or a recommendation to investors or potential investors in relation to holding, purchasing or selling units in the fund and does not take into account an investor’s particular investment objectives, financial situation or needs. Before acting on any information, investors should consider the appropriateness of the information having regard to these matters, any relevant Product Disclosure Statement (‘PDS’) and in particular, they should seek independent financial advice. For further information and before investing, please read the PDS available upon request and Target Market Determination (‘TMD’) located at https://www.channelcapital.com.au/design-and-distribution-obligations.
The Fund as described in this document is not for sale to US persons and no offer is being made to US persons.
