There is no instance of a nation benefitting from prolonged warfare.
Sun Tzu, The Art of War
Note: None of the material or content contained herein constitutes investment advice. It is solely for educational and informational purposes. Past performance is not an indicator of future performance.
Nvidia’s market cap currently sits at 5.46T, larger than the annual GDP of every country in the world except the US and China and larger than all the markets in the world except US, China and Japan. Microsoft has a market cap of over $3 trillion, Apple $4.3T and Google $4.8T. AI has driven incredible amounts of paper value on these stocks alongside very real capex expenses.
Source: Apollo Global Management, Putting the total amount of hyperscaler capex into perspective Feb 2026
It is a gold rush, some companies are buying mines and some are selling picks and shovels. Nvidia are in the envious position of selling the picks and shovels and even more enviable is that they sell the best picks and best shovels and everyone wants the best. They have more control over their outcomes than those buying mines and hoping to find gold.
The war is happening between the miners, the hyperscalers, Microsoft, Google, Meta, Amazon, IBM, Oracle and increasingly OpenAI and Anthropic. They see a golden future, they’re not quite sure how much gold there could be but they are confident there should be gold on the other side. They are unlikely to all be right, some mines have gold and some don’t but people are excited and they don’t want to miss out they wager it’s better to back anything than nothing. This is not a good strategy.
Sam Altman recently said, “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes. Is AI the most important thing to happen in a very long time? My opinion is also yes.
In our recent BREAKINGWATERS Boom and Bust conversation on bubbles, it was clear that the AI rush closely resembles a bubble but, as with all bubbles, it has it’s own unique characteristics. William Quinn, co-author of Boom and Bust and, senior lecturer on bubbles, noted that one of the key differences here is that the capital in this instance is not coming from the everyday investors. It is mostly a mix of corporate bonds, private credit and sovereign fund capital being deployed to the cause. A lot is off balance sheet too. Professor John Turner, the other co-author noted that this may also be a politically driven bubble masquerading as a technology bubble - when Deepseek burst onto the scene with a step change in performance for a fraction of the cost, many have declared we are in an AI cold war - see this headline from the Washington Post a few days ago.
This war is putting huge pressure on the players involved. Cash is being hoovered up, debts levels are skyrocketing and all in the name of, at this time, unclear investor value creation (there is clearly value creation to consumers). There is no doubt that revenues are rising fast but this is not a fait accompli for success for investors.
AI is starting to look like it will be a utility for life, like the internet or smartphones, it will become a necessity of the everyday for everyone. Utilities though tend not to be very attractive investments. For the internet and smartphones cycles, the value went to those that built on top of the infrastructure. Apple was an exception in the amount of value it could extract from it’s hardware, a result of it’s magical marketing (see this Sun Tzu-esday) but in general the ones that made the most money were software, search, social media and market places. Low capex, high profit.
In this new cycle we are already seeing some bifurcation of roles. Anthropic are leaning towards off the shelf enterprise solutions by passing a host of specialist providers that have popped up. Open AI is where everyone started but, if my anecdotal conversations are right, people are drifting and trying other models. Most seem to land on Anthropic’s Claude or Gemini or a combination of both. I use Gemini I have never had no reason to try anything else. I have used ChatGPT and it wasn’t as good, it gave so so answers or at times it just couldn’t deliver an answer, something I have never experienced on Gemini (an indication of Google’s infrastructural advantage). The other options are Chinese AI providers and Tencent, Alibaba and others are in their own regional war. They are tending towards a lower cost or in some cases free model. Deepseek have aimed for a cost efficient offering too.
These hyperscalers must continue to spend to try eek out a lead and that is their primary focus in this phase, grab share and worry about everything else after that. Making money is a problem to be figured out once they have won the space. This clip from Silicon Valley gives an insight into how the thinking works in Silicon Valley.
The longer it goes on the more damaging it will be to the long term prospects of the “nations” involved as they burn ever more capital in pursuit of the goal. It has the hallmarks of the telecoms arms race which ended with a slew of lumbering debt laden utility-like giants that were unattractive investments and ultimately created little value for investors. Absolutely, not all players will survive this build out, they will either be wiped out altogether - I am thinking of the more fragile businesses who need more pieces to deliver the goal (capital, infrastructure, distribution) or they will be severely damaged and their hands tied with debt like the once dominate telecoms.
I have no doubt that there is a transition started in the global order of companies. Every couple of decades, new businesses rise up to dominate for a long cycle and I think we are at the beginning of this. There will be a business that unlocks a use case with AI that is low capex, high take up, high profit with a flywheel that will roll them into the future unlocking further profit channels. I don’t know what that is yet but I am watching. It’s important too to realise you don’t have to see it before anyone else for it to work, you can still make a lot of money if you miss the first rally, there is time - just listen to Peter Lynch:
As I look across the landscape, I do have thoughts on what I am seeing - paid subscribers only
