A Warning: The AI Arms Race. DeepSeek and Black Swans. China, The US, and Taiwan.

“Taiwan is really the engine that’s driving AI,” said Sean King, a senior vice president at Park Strategies and a former US Department of Commerce official.

The future of everything is uncertain.

Alright, back to the geopolitics.

This week the new Chinese AI app DeepSeek reached the number 1 spot in the IOS app store charts. DeepSeek is open source, meaning the code is open to the public and entirely free. If you want, you can download the code base and run your very own GPT on a local machine, and it rivals (and in most ways, exceeds) its competitors, like OpenAI's ChatGPT and Google's Gemini, both of which have proprietary code and crucially, they charge for use of their advanced services.

As a matter of comparison, this would be like opening a bakery specializing in custom organic pastries with trademarked recipes staffed by world class bakers, just to have 3 guys open an even better shop next door selling everything for free.

A lot of people are saying this is a big "one up" for China–and it is–but really this is a battle between closed source and open source code. When it comes to machine learning, open source moves much quicker. But this does undermine every major AI software developer in the united states, and it remains to be seen how big the fall out will be.

To be clear, the development of DeepSeek is nothing short of incredible. They accomplished all that their competitors have with much less.

Black Swans

The expression "black swan" was used to represent things which do not exist, until 1697 when Dutch Mariners saw black swans living in Australia, at which point the expression was reinterpreted to mean an unforeseen event.

The reinterpreted theory was developed by Nassim Nicholas Taleb in his book "black swan", to explain:

1. The disproportionate role of high-profile, hard-to-predict, and rare events that are beyond the realm of normal expectations in history, science, finance, and technology.
2. The non-computability of the probability of consequential rare events using scientific methods (owing to the very nature of small probabilities).
3. The psychological biases that blind people, both individually and collectively, to uncertainty and to the substantial role of rare events in historical affairs.

Maybe a bit wordy, but here is the point: DeepSeek was a black swan for the vast majority of investors and Tech CEO's. the likelihood of an Open Source AI being released to the public was not on the 2025 bingo card, especially not one that is claiming to be cheaper than its competitors while still being better in almost every way, and it has major implications for the world we live in, and the future of technology.

The reason I bring up black swans though, is for a brief tangent on investments.

Us Tech stocks took a steep hit this week once DeepSeek was released, as did crypto largely.

When you're looking at trading prices, trying to make investment decisions based on past performance, you are not factoring in the likelihood of black swans (by definition, you cannot factor them in). this is why I am an advocate of long term investing decisions. Short term trading may work out for you, for a while; but all it takes is a string of bad trades to upend your profits.

Elon musk can tweet something and destroy months of profit. Trump can sign an executive order. China can release an AI. A large crypto exchange could be hacked.
These events are unpredictable in nature, don't hedge bets on everything going according to plan (it rarely does). If you read my series on Crypto and the changing world order you should be equipped to make longer term decisions, and hedge risks. When you have a 10-20 year time horizon, these bumps along the way get ironed out (usually).

Taiwan

In the 1990's the US produced 40% of all semiconductor chips (the kind that power computers, phones, and AI machines). Today we produce under 12%, and Taiwan produces 60%. Taiwan produces 90% of the most advanced chips we have today, and the US produces none.

These chips aren't just needed for consumer goods, they are also used for military applications, and businesses running data centers (Netflix, Facebook, youtube, etc.) and of course AI machines (OpenAI, Google, Meta, etc.)

Brief Disclaimer: I am going to be talking about US presidents. I am not making a political stance here. Even the best president will make bad decisions, and a broken clock is right twice a day. I am only concerned with facts, and their consequences.

In 2022, then president Biden launched the CHIPS act to bring semiconductor manufacturing back to the US, allocating $53b of incentives to stimulate US manufacturing and R&D.

The goal is to get more US companies producing semi-conductors instead of outsourcing to Taiwan. Chinese relations are tense over Taiwan, as they want to reunify the Island into the mainland; and US reliance on the Island has turned the nation into a proxy of both Chinese and US interests. If ever there were a place for WWIII to kick into high drive, it would probably be over the future of this Island.

What does this all mean? more than I can say here. but to (over)simplify:

1. US semiconductor manufacturing (and R&D) is going to take time to become competitive with Taiwanese manufacturing (and R&D). A CHIPS act II and III will likely be needed to bring the US up to par. $53b is surprisingly not enough money to get the US back in the manufacturing game.
2. China has a vested interest in reducing US reliance, as has been made clear with the BRICS nations goal to stop using the US dollar as a reserve currency. China's economy is still reliant on the US for consumer spending (their GDP is made up almost entirely of exports), but if they become the global leader in semiconductor manufacturing, this could turn the tides (somewhat) in their favor. Though... a lot of this remains clouded in mystery (remember the black swans from earlier?).
3. Taiwans economy is extremely reliant on Chinese trade, though it fell to its lowest level in 2024. The CHIPS act included stimulus to help Taiwan reduce reliance on China by allowing them to build semi-conductor manufacturing plants in the US (building locations in Phoenix Arizona)

Oh, and remember when I said I am not taking a political stance? please remember that as we move forward.

Enter: Trump

To bring chip manufacturing back to the US we need even more stimulus. So will Trump enact a CHIPS II and III?

Well... no. Or, maybe?

He has criticized the act leading up to the election and said that he plans to repeal the act, but also either contradicted himself or changed his mind, later noting that he did not have plans to repeal the act.

Trump's nominee for secretary of commerce, Howard Lutnick noted in a staff meeting that he planned to move forward with the CHIPS act.

Trump has also said "That chip deal is so bad." criticizing sending billions of taxpayer dollars to "rich companies" and suggested imposing tariffs on foreign-made chips would be a better way to move production to the U.S. Whether it was a bad deal or not is to be determined in the eye of the beholder, but he did specify that "they would not send us the good companies anyway", which is a bit strange given that TSMC (the Taiwanese company who built a plant in Phoenix) is the largest semiconductor company on the planet (by revenue anyway).

As usual with US policy, there are countless objective and agenda's strung together into each proposal, usually including some the trojan horses of some partisan efforts. the specifics of the deals and how they are packaged is a subject for another time, so here I am just focusing on this individual goal.

To be clear, I am not broadly criticizing any US president here including Trump. I am very narrowly focusing on Semi-conductors, which is barely scratching the surface of presidential duties.

It's unclear at this point if Trump plans to continue the CHIPS act and impose tariffs on foreign semiconductors, or just choose one or the other. It's also unclear if he is threatening tariffs, or actually plans to impose them. Trump has a history of threatening tariffs to get his way, Columbia recently bent the knee when he threatened to impose tariffs on them. When other countries keep believing his threats, it's impossible to tell if they are empty or not.

One thing has been made abundantly clear though: Tariffs are great in theory, but less than great when they touch the bedrock of reality. For example, when Trump imposed tariffs on foreign washing machines in 2018, not only did the prices of foreign washing machines (like LG and Samsung) go up, but domestic products increased in price as well simply because they could.

It's a rare event (sans a domestic catastrophe) that a consumer goes out to buy a washing machine without a matching dryer. Because they are usually sold in pairs, the price of dryers went up across the board too. Again... simply because they could. No tariff was placed on dryers.

The strategy of tariffing foreign semi-conductors would very likely have a major impact on the cost of semi-conductors manufactured in the US. And DeepSeek seems like a response here. China plans to continue their advancements in AI software to become the global leader in it's technology, and they plan on unifying with Taiwan on a longer timeline.

The US response so far seems to be... complicated. But it involves reducing our dependence on foreign semi-conductors (recently through CHIPS and by having TSMC build manufacturing plants in Phoenix, in the future... we do not know), and this has bipartisan support. the method of achieving this goal is where we have a partisan gap.

My own thoughts are again, not politically motivated. Tariffs make great threats but worse promises, and the effect of raising the cost of semiconductors but only for America is a likely outcome of tariffs, and this seems like a losing strategy.

But in all likelihood, tariffs are a really bad idea. The importer, not the exporter is who pays the tariffs. This would likely lead to a mass exodus of big tech companies from the US because of rising costs. This would be horrible for the US economy, job market, the future of US innovation, and our daily lives.

Time will tell exactly what Trumps plans are, and how those plans shake out in reality.

AI and Crypto

So what does this all mean for Bitcoin?

It means a few things. First off, the US dominance in AI software seems to be softening, though time will tell what the future looks like. Open source software moves quickly, and AI moves quicker still.

US Tech companies have relied on proprietary code (closed source code) to increase marketshare. The benefit here of course is funding, and being able to guide the AI toward a particular goal, however the company behind DeepSeek claims they trained their AI model with $5.8m of resources, compared to $17.9b dollars for OpenAI (so far).

They also claimed (crucially) that they trained the models on significantly fewer chips than their US counterparts, hence the stock drop for Nvidia and other chips manufacturers. This claim has been scrutinized but if it does turn out to be true, this would be bad news for US tech companies. If it turns out not to be true, we should see a rebound in stocks and crypto.

What this means, is that developers are getting very creative and breaking previously held dogmas. Remember Eric Liddell flailing his arms and bolting off the starting line, and doing what no one thought was possible? that pales in comparison to the kind of ingenuity we are seeing in the AI space right now. It's closer to splitting the atom or the first airplanes.

Decrypting Crypto

Disclaimer: This part is still "science fiction". My point in bringing it up is purely to discuss how far we have come with regard to technology, and how many of our dearly held electronics can become obsolete at a moments notice. I am not fear mongering, it's just a conversation worth having. please do not think the following is an accurate prediction of what will happen, but rather a philosophical point. how do you operate your investment portfolio in light of great uncertainty? that's a question you should be seriously pondering.

Making the code open source speeds up development, but it does pose certain risks.

I don't want to be sensationalist here, because we are still quite a ways off from this (and it's unlikely to happen), but:
Rapid development of new AI software, rapid development of new semi-conductors and Googles new Quantum chip for example, could lead to a world where encryption is a thing of the past; or at the very least needs a massive overhaul.

Encryption (like the "crypto"graphy of cryptocurrency) are what keep it extremely secure. But it does not start and end with Bitcoin, encryption is necessary for banking, military secrets, healthcare industries, and many other aspects of our every day lives.

Black swan events like a 10x cheaper and 10x more powerful AI may seem trivial, but these types of things could be the canary in the coal mine for a massive increase in computing power, on a level which could unravel life as we know it.

Open source models are just the beginning, where we really kick off is in the development of super computer chips which–cannot currently–but eventually will be able to decrypt crypto.

Now thankfully, we are still a ways off from this, and the entire world has much more to lose than just Bitcoin wallets in the face off decryption (which means major interests are making sure this does not happen... or are trying).

But I end with this note to suggest (or re-suggest) that we should all be very wary of making predictions of the future. New AI and semi-conductor advancements are just the beginning. If you look at the timeline of AI development vs. the early 2000's .com boom, we may not have even approached Y2k.

Light at the end of the Tunnel

As always, some brief consoling.

One very likely outcome of the increase in semi-conductor efficiency, and software efficiency, is an improvement in crypto mining and staking efficiency. Mining Bitcoin is notoriously expensive and energy intensive. Finding shortcuts for AI algorithms and new developments in chip manufacturing will increase the output of mining machines (and other crypto related paraphernalia) while reducing its carbon footprint and energy expense.

These are all very good things for crypto, and for out daily lives. Much online rhetoric about AI is negative (outside of the silicon valley twitter space), but it is not all doom and gloom and paperclip maximizers.

So what do we do?
In my opinion there are a few things:

1. Investing in crypto still gives you options, and I genuinely do not think decryption is a real risk on any kind of time horizon that makes sense. Super computers are vastly more expensive to run than a standard datacenter (which is already absurd) making it highly unlikely someone outside of a well funded research lab will be able to operate one. This does pose a risk for future developers, not so much for the average investor today, and I will be looking into advancements in encryption technology as time goes on.
2. Tech stocks are insanely volatile right now, diversification is key here.
   it's going to be very hard to pick winning stocks going forward, don't keep your eggs in one basket.
3. AI crypto projects are a great way to invest into the AI space, especially outside of the US. While US residents can invest in foreign stocks, the requirements most of the time make it unfeasible for the average person. this are going to be volatile as well, exercise caution and read my series first.
4. Non-tech/crypto assets are also a wise decision for diversification sake. regardless of AI development, people are going to need houses, food, healthcare, etc.

Hope this one found you well too.