Weaponisation of AI (stuck in a prisoner's dilemma)
Weaponisation of AI (stuck in a prisoner's dilemma)
The fight for global dominance and corporate survival. Game theory and the law of unintended consequences. Tit for Tat and a race to the bottom.
Remember the price spike in Oil & Gas that occurred after the Russian invasion into Ukraine?
The narrative was that Putin was weaponising energy — making it immediately more costly and precious. At the same time, the EU decided to reduce dependence to Russian energy as retaliation for the invasion.
During the summer of 2022, WTI Crude passed $120 / barrel and Eurogas (Dutch TTF) hit €350 /MWh. By September of 2022, the Nord Streams had been hijacked and “mysteriously” blown up.
Could anything else end up blowing up? Maybe Taiwan? What about Big Tech? Let’s investigate.
Note: WTI is now barely at $80 and Eurogas is 10X lower…
Today we will explore the setup in the AI/Semiconductor complex and its geopolitical dynamics. We will also touch upon the weaponisation of AI from tech cos and the effects and consequences that I expect from this emerging trend.
Two years ago I sent out another post on the East VS West / Cold War theme which focused on the financial ramifications of the Russo-Ukrainian war.
Having written it 2 years and 2 weeks ago, I think I pretty much nailed it.
Namely:
The Euro/USD did test and break parity 6 months after sending.
The EU is the trading bloc that is suffering the most from this.
The USA is doing “fine”.
We are experiencing a sort-of globalised protectionism and a general trend of de-globalisation.
Ukraine got wrecked.
Inflation is persistent and real assets have been performing the best.
And more… read it below.
SUPREMACY
Military. Monetary. Economic. Technological.
All these four classes of supremacy feed into each other, in ways that’s difficult to fathom — and yes, they do overlap.
But where does AI fit in all this? Is AI important enough to be placed in the “Supremacy” bucket? I don’t know but it seems everyone else does — if they didn’t we wouldn’t be having an AI race right now.
Ultimately, I don’t think AI supremacy is a category on its own — it just falls under Technological Supremacy.
East Vs West Kicks Off
In January of 2018, DJT set trade tariffs and barriers to China as retaliation to unfair trade practices and intellectual property theft. China responded with its own retaliations and the US Vs China trade war was in full swing.
Since then, the US started to take increasing measures against an ascending China as they saw Chinese technology companies growing outside Asia. The Americans knew their global hegemony was partly dependent on a continued global domination of Big Tech (you know, iPhones, Amazon and Google search everywhere). The Chinese knew it too, and that’s why they are trying to compete.
The story evolved when in 2019, Washington blocked US companies from working with Chinese networking and mobile telephony company Huawei. The accusation was that Huawei’s technology could be used for espionage — and therefore a threat against U.S. National Security.
The Washington Vs Huawei attack was launched just in-time to stop Huawei selling their 5G technology to Western countries looking to upgrade their infrastructure. Huawei’s smartphone sales are also suffering internationally as a consequence of this.
The West wasn’t ready to allow a Chinese company to sell tech to Western corporations and households.
The battle then moved to the Semiconductor front…
In September of 2020, the US-led restrictions expanded and now focused on the Chinese semiconductor industry — the objective was to choke the Chinese from advanced tech and prevent them from further developing their semiconductor industry. The narrative from the US is that they don’t want China getting its hands on technology that could be used against the US.
The restrictions first focused on SMIC, but by October of 2022 expanded to the totality of the Chinese semiconductor sector with the sole intention of completely choking it off from advanced technology.
Supremacy & Dependence
Their motivation seems to be not only economic but also military.
On economic supremacy, letting China freely implement cutting-edge technology means the West has no edge against them and that they can compete against the West in the global technology space. The US prefers to hold all the cards and keep China dependent on them — but cornering China too much means things could boomerang!
Remember the sanctions against Russia? How did that work out?
On military supremacy, an eventual US Vs China military conflict means the US needs to stack up advantages against their potential enemy and prevent it from obtaining the same technology and capability.
As an example, Palantir Technologies (warning: overvalued cult stock!) won contracts from the US Army to develop AI-led technologies that could offer an advantage in the battlefield.
But let’s get to the point…
China got the message
The growth driver for China in the past ~20 years was infrastructure investment and real estate. This required a lot of energy and commodities —benefitting the commodity trade and its derivatives in the process (and a lot of shipowners!).
Now, with the Chinese economy having piled on too much debt, the Chinese feel it’s time to step back from that trade and focus on the semiconductor complex and technological supremacy in general.
In February, it was announced that a $27bln fund is being raised to invest in cutting-edge technologies to antagonise the US-led sanctions. For the moment, China can’t procure EUV lithography machines from the world’s sole supplier (ASLM in the Netherlands) — but I reckon the Americans will push them to eventually build the own.
Hyperscalers and Corporate Survival
Last week I wrote about Hyperscalers and Microsoft’s battle to not lose the Cloud race too.
How is Microsoft looking to gain an edge?
They are trying to bring in-house the best AI tech and integrate it to all their offerings to create an unbridgeable edge between them and the competition (namely Amazon and Google).
The investment into Open AI (covered in my Nvidia piece here) and the integration of GPT-4 with the Microsoft suite of products (Cloud, Office, etc.) means Microsoft has an edge Vs Google and Amazon which don’t have a deal with Open AI — or their own operating AI models ready to be used.
Last week, Microsoft also struck a deal with Inflection AI by poaching its top staff and licensing its technology. This, by the way, was not an outright acquisition but something else..
What did Microsoft do here?
They poached Inflection’s top brass and talent, and paid Inflection $30mln to not sue them back for poaching said talent.
They paid Inflection $620mln for licensing its technology in a “non-exclusive”arrangement.
And somehow they found a way to “convince” what remains of Inflection to pivot away from building an AI chatbot and become an AI “studio” helping other companies work with LLM (large-language model) AI.
Cool story bro…
The Antitrust Stage
Microsoft is not a stranger to antitrust troubles. Right now Apple, Amazon and Google are facing antitrust lawsuits — making Microsoft realise it can’t outright purchase all the AI tech it wants to hoard for itself; else the regulator will come a knocking (again).
This is why they aren’t actually acquiring their companies, but finding solutions to bring them in-house while closing all paths for other companies to work with them.
Do I think that they will eventually awaken the bear (FTC & DOJ) and get sued? Yes!
So is this the wrong play by Microsoft? No!
Game Theory and the Prisoner’s Dilemma
Well, what does game theory say? The most famous example of game theory is the prisoner’s dilemma. Microsoft (and Big Tech in general) is trapped in a sort of prisoner’s dilemma here.
If Microsoft doesn’t push the limits of fair competition, it will never gain an edge from the competition — and unlike Microsoft’s other businesses, the Cloud is a very capital intensive business. Microsoft needs an edge to dominate this, and it’s obvious that the edge has to come from AI.
By the time the regulators catch up to Microsoft, the latter would be well on its way to not only in-house other AI models (e.g. Inflection, obviously) but develop their own internal AI models while starving the competition from the models they acquired.
This game plan simply makes overwhelming sense, from a game theory perspective.
TIT FOR TAT
What will the others do?
But Microsoft isn’t only racing against regulators, it’s racing against other tech players. Apple is already in discussions with Google to integrate Gemini AI to its ecosystem, and put some wind under the sails of Google’s AI ambitions. Apple has been working on its own AI LLM as well, so don’t you worry.
But there’s more, this would mark a broader strategy of co-operation between Apple and Google — which could potentially have negative repercussions for Microsoft.
Microsoft envisions becoming a serious competitor to Google search through its now AI-powered Bing — but Apple now owns 10% of the worldwide computer market (and 16% of the US one). Remember when Microsoft was alone? Well….
With the reputation that Apple has — launching a cheaper version of the Mac would result in quickly expanding market share and an erosion in the Microsoft moat. The outcome from this multi-front attack is Microsoft goes to win in one segment while getting killed in another.
And if Microsoft is allowed to tie-up Microsoft CoPilot (its AI productivity tool), with its own search engine (i.e. Microsoft Bing) — then why can’t Apple tie-up its own ecosystem with the Google search engine?
Insert lightbulb moment!
The Law of Unintended Consequences & Fat Tails
Chips as a commodity
Dominance by NVIDIA in GPUs means custom silicon is becoming the norm, creating the chips-as-a-commodity dynamic.
That’s not what Nvidia intended, and I believe that’s why it’s funding and pushing Coreweave so much — to compete with the big 3 hyperscalers which are all creating their own silicon.
Cyclicality in semiconductor margins
The US Chips & Sciences act aims to help the US catch up in the AI/Semiconductor race — but it will probably end up creating excess fab capacity, that will result in a reduction in pricing.
Intel has a $100bln investment budget for new fab capacity over 5 years. Imagine if that doesn’t work out as expected!
The Chips Act from the US is a top-down policy that usually comes with unintended consequences. We will keep following this theme to find out what they are and their full effect — for now I see excess fab capacity down the line, inflation and market distortion.
Chinese antifragility? Taiwan blockade?
The $27bln Chip Fund from China will result in a tech/AI race from the Chinese to catch up to the West. This could result in corporate espionage, intellectual property theft (of the highest order) and/or even a massive pivot by the Chinese to becoming a nation of much higher-quality technological development, down the line.
If the Chinese aren’t successful in creating their own 7nm nodes and be at par with the West, they will start causing grief to Taiwan. This could even include an all-out invasion if the West manages to squeeze China too far.
The whole world relies on advanced-level foundry capacity from Taiwan, which sits at >80% of the global total. And this is why the US is pushing so hard with the Chips & Sciences act — to bring capacity back home and robustify itself from any problems in Taiwan.
A Race to the Bottom
As I finish off writing this piece — Nasdaq futures are down 60bps.
Google, Intel and Qualcomm executives say they plan to battle Nvidia’s AI dominance. This is the over-arching theme in modern tech: all-out competition, creative destruction and a race to the bottom. Big tech is shifting from growth and profitability to fierce competition and cannibalisation.
I will stay all over this theme (and others) and continue to offer you generalist and multi-thematic actionable insights that I think are hard to find anywhere else.
Sincerely,
Philo