Chinese AI is entering Western markets in a way that can only be described as an โAI invasion.โ
As investors sell Western tech stocks because of DeepSeek, they face a new reality. Selling NVIDIA or ASML effectively means punishing Western firms over Chinese innovation.
This is a sign of how geopolitics and the world of investment are becoming interwoven in new ways. The same investors who were pressuring US companies to leave Russia for ideological reasons or want to reposition their portfolio to support America First, must now also approach DeepSeek through a similar lens.
Is selling Western stocks in the interest of the West, since it may weaken US and European tech firms and since Chinese financiers may not be tightening their purse strings towards Chinese Big Tech or BAIT (the Chinese equivalent to FAANG)? Selling stocks is no longer just about punishing a company over performance or (lack of) innovation. It is also about investors acknowledging that the US-China fight has entered a new phase, where the Western response to Chinese innovations, including in stock markets, will shift the balance of power.
However, this is only one part of a much bigger picture.
๐ฅณ๐ฎBig Announcement: 2025 Geopolitical Futurecast Now Available to Subscribers!

DeepSeek has disrupted the state of play between the worldโs two largest economies by introducing AI systems that outcompete Western equivalents at a fraction of the cost. The China-based firm has achieved this even with US export controls in place. These controls were supposed to restrict Beijingโs access to cutting-edge technology and, in the process, slow (or paralyze) Chinaโs rise with AI. The opposite appears to be happening.
The world is back to viewing China as a โclosed safe,โ where nobody really knows how advanced China is in terms of technology or military.
Just one aspect captures how the AI battlefield has changed in months:
In just three months since November, DeepSeek has launched several different AI models, showing the scale and speed that the company is working with. Chinese AI is entering Western markets in a way that can only be described as an โAI invasion.โ
In November, DeepSeek launched R1-Lite. In December, it launched V3. In January, it launched R1. And on January 27, as the global stock sell-off accelerated, DeepSeek unveiled โJanus-Pro-7B,โ effectively its own AI assistant that can rival popular GenAI services like DALL-E 3. Connected to all this, DeepSeekโs models are, on average, 95% cheaper than Western models. DeepSeek may have just 50,000 GPUs to train its models. Compare this to Microsoft, which wanted 1.8 million GPUs by the end of 2024, or Meta, which is dishing out a total of $30 billion to increase its GPU size to over 1 million. In comparison, the US government spent $28 billion to put a man on the Moon.
With fewer GPUs and less overall cost, DeepSeek produced AI systems that are similar to much of what the US offers.
The question spinning Silicon Valley: How did this even happen?
The question that should be spinning Washington: What does this mean for America?
DeepSeekโs new systems signal a turning point in geopolitics where Chinese systems have (almost) leaped ahead of America.
DeepSeekโs systems muddy other predictions about Chinaโs leadership gap, such as that by 2032, China will only produce 2% of advanced global chips. The reality is that all certainty about China is gone (again). The world is back to viewing China as a โclosed safe,โ where nobody really knows how advanced China is in terms of technology or military.
As the world takes stock of DeepSeek, geopolitics is front and center, meaning this is not a technological disruption and then a geopolitical disruption. It is geopolitical first. From stock sell-offs to user adoption, the world must look at the Chinese AI outfit for what it is: a sign that China is no longer a rising power; it has risen, and it is now able to match or even exceed America in certain domains.
Are alarm bells ringing?
WESTERN MARKETS FACE AI DISRUPTION
The models that DeepSeek is injecting into Western markets represent a โdangerous momentโ for Western governments. For years, from Washington to London to Paris, the West has been investing tens of billions to drive critical industries forward, none more important than AI sectorsโincluding chip development.
At the center of this strategy was building up a new economic engine.
Large Western technology firms would build advanced commercial AI services that tens of millions of consumers use (and pay for)โan AI loop of supply and demand. However, DeepSeek puts the entire strategy of Western states and businesses at risk. If Chinese AI firms can offer the same services as Western AI firms at a fraction of the cost or for free, where will Western consumers go?
This is not just a business emergency for OpenAI, Amazon, Microsoft, Meta, or Apple. It is also a โgeopolitical emergencyโ for Western governments. Their economic vision is being sidelined by a new set of Chinese exports, namely advanced and cheap AI platforms.
Is this a deliberate strategy by China, even a warning shot to Trump?
Strangely, this is not a new Chinese strategy.
The entire argument for restricting Chinese imports and going after Chinese overcapacity is based on Chinese goods undercutting local competition, largely because of state subsidies from Beijing. The Chinese can produce and sell at a price point others cannot. And while before, this centered entirely around steel or EVs, now it extends to AI.
Equally important is that DeepSeekโs models represent the first real wave of Chinese AI in the Western world.
The West does not have experience with Chinese AI the same way China has experience with American AI. Even OpenAI, which skipped China when launching ChatGPT, was still pulled into the Chinese market, as Chinese users, hungry for American innovations, used VPNs to access the AI service. In the end, China went after its large tech firms, forcing them to restrict access to ChatGPT as Beijing viewed ChatGPTโs answers as โAmerican propaganda.โ
Now, however, the West is grappling with DeepSeek, which is now the number one downloaded app in the US App Store (Apple). How will Western governments react to this kind of Chinese advancement (across the border), threatening their economic design and business success?
THE RISE OF โGREY BUSINESSESโ
Somehow, DeepSeek was able to build a massively complex AI system even though US sanctions were in place.
There are two possibilities.
Either DeepSeek trained on Chinese-built hardware, part of the Chinese self-reliance drive to replace US software and technology with locally developed alternatives, for example, Huaweiโs Ascend 910C chip, which is entering mass-production, especially as ByteDance orders 100,000 units. Or, DeepSeek used American blacklisted technology, like Nvidiaโs advanced AI chips like H100 or A100, the former of which was never sold in the Chinese market, and the latter of which was sold until 2022, and which costs $20,000 a piece on the black market, double their price in the US.
Even with US sanctions, Nvidia still controls around 80% of Chinaโs AI chip market. In fact, in 2024, even with US trade limits in place, Nvidia was expected to make upwards of $12 billion from chip sales in China.
Consider what this means.
A company that could soon become the most important Chinese AI firm in modern times built its flagship systems by evading and bypassing US sanctions. Some may view this as breaking the law. But, as millions upon millions of people adopt R1 (and the next systems DeepSeek launches), this will become the accepted reality. The need will outweigh infractions.
A new normal has been established, where companies can operate in the grey, bypassing sanctions and building products that are successful in global markets. If DeepSeek can do this, it will motivate others to do the same. The world may be about to witness the birth of new companies, from China to Russia, that evade US sanctions and become global commercial success stories.
In geopolitics, the lines separating right and wrong, legal and illegal, are beginning to blur even more.
AMERICAโS GEOPOLITICS GETS EXPOSED
As lawmakers in Washington make sense of DeepSeek, they will see a separate message on the wall: the US sanctions no longer work. The teeth and strength of US sanctions are on full display.
The ineffectiveness of sanctions is not new. The multiple sanctions packages the West has imposed on Russia, the decision to go after โthird partyโ nations, and the new US โglobal export frameworkโ all point to US sanctions not being able to meet their intended objective.
What is looming now is a complete rewrite in the US as to what kind of sanctions should be imposed and how far-reaching they should go. The US cannot afford to have nations question the reach and power of its sanctions. Once the ball starts in this direction, there is no going back.
This is not just about Chinese AI.
US sanctions are about โfreezing nationsโ in place until they change behavior and adopt a new course of action.
Alongside China, there is Iran, which is still moving towards becoming a nuclear-armed state. If China can unveil unbelievably powerful AI, and if Iran tomorrow becomes a nuclear power, all while US sanctions are intact, few will fear Washingtonโs sanctions wrath.
If existing sanctions no longer work, the US will have to either shift gears and focus on trade, commerce, and broader geoeconomics or consider new angles.
Towards the end of the Biden administration, the White House proposed โtweakingโ the AI weights in US AI systems. This would change how AI systems reason and โdo somethingโ (i.e. generate an answer, produce an image, etc.).
If this were to be enforced, it would completely change the game, as US AI exports would have to meet certain โgeopolitical conditionsโ1 before they can be sold around the world.
With sanctions in the red zone, new radical ideas could soon be implemented, once again shaking the waters on the world stage.
WHO CARES ABOUT CENSORSHIP?
A big part of the fears around DeepSeek concerns censorship. This should not be viewed as insignificant or business as usual. DeepSeekโs models appear to come with โbuilt-in censorship.โ
For example, look at the following video posted on X:
The user is asking DeepSeek about a โfamous pictureโ where a man stands with โgrocery bags in front of tanks.โ This is, of course, the infamous picture taken in 1989 at Tiananmen Square, also known as โTank Man.โ
In one of the last frames, DeepSeek answers the question before scrubbing its own answer (see below) and replacing it with an automated answer, either an inability to comprehend the promptโor censorship.
Clearly, the rules and beliefs of the Chinese government are โhardwiredโ in DeepSeekโs systems. Also, considering that large tech companies have to prove to Chinese regulators that their AI embodies โsocialist values,โ DeepSeekโs AI may be expanding Chinese ideology into markets everywhere, injecting new parameters and limits for users.
This is not just about the West. In the US, people might not be able to get accurate information about Tiananmen Square. In India, it might Tibet. In Japan, it might be the Senkaku Islands. In Russia, it might be the Bolshoi Ussuriysky Island. The Chinese geopolitical position may be reflected through DeepSeek (and other AI) models, influencing millions outside of China and colliding with the positions of foreign governments.
The conditions are forming for China to collide with several nations who view DeepSeekโs systems as effectively a megaphone for the Chinese state in their sovereign societies. These governments may ask for the censorship to be removed or else.
However, there is a separate question here, one that has not been asked until now. Who cares about censorship?
If tens of millions of people are able to access cutting-edge AI for free/cheap, they may not care whether they can get an answer about Tiananmen Square or Tibet. They want to generate an image, produce creative text, generate new video footage, or parse through complex code. The geopolitics is likely to be far from their focus, if even considered.
This may be exactly what China is banking on: a consumerโs โhunger for innovationโ versus their โappetite for geopolitics.โ
This puts Western governments in an extremely difficult position, which could be described as a โTikTok position.โ
Just as TikTok has created a new challenge where consumers/citizens want the platform for their economic security but governments want to ban it for national security, so too could Chinese AI systems deepen these fault lines. Do Western governments put productivity first (keeping the Chinese AI systems online) or geopolitics first (taking the Chinese AI systems offline)?
THE COMING RETALIATION
For those looking at the bigger picture, it is clear what direction this train is moving in. The US and the broader West cannot allow such AI systems to exist unrestricted in their societies. They challenge the competitiveness of businesses and the nation.
The Stargate project, a $500 billion strategy unveiled by Trump just as DeepSeekโs AI started to make headlines, seeks to build out a huge network of data centers across the US to power AI. The entire project could create over 100,000 jobs spread across several US states. However, the entire premise of Stargate is two-fold: first, American AI services will become more popular and task-intensive, and second, AI needs a huge infrastructure to operate.
Models like R1 change these ideas. DeepSeek built its AI and operates it at a fraction of the cost of Western companies. Americans may very well flock to R1 (and the next AI models) over US offerings purely based on price point. This puts a project that is two-thirds the budget of the US military or one-third of the Canadian GDP in limbo.
The US and its allies cannot afford for this to happen. Chinese AI cannot disrupt their economies. Therefore, these governments are likely to limit or ban such models in the near future. Whether on the grounds of national security or through โdivest or banโ type bills, the West will not allow Chinese AI models to operate according to free market principles. This could end up defeating Western companies in their home courts.
Such a notion is extremely scary and provocative.
This means that business competition is now being defined by geopolitics. Firms will exist or disappear not because of innovation, consumer appeal, or price point, which are the main business calculations, but because governments enforce their borders and establish new demarcation lines.
Once again, geopolitics and business are converging.
ACTIONABLE TAKEAWAYS
The US-China fight has reached a critical turning point. All ideas and broader predictions about the US-China rivalry need to be reexamined. How far behind is China? How far ahead is America? Whatever certainties existed are gone. DeepSeek is a reflection of the โtrue Chinaโ that exists beneath the illustration drawn in the West.
Copying has evolved into something more. With the exception of Chinaโs green industry, Chinese innovation has largely revolved around the US for some time now. But with DeepSeek, China is using the same โfundamentalsโ as the US but creating something superior. In fact, in the case of DeepSeek, China is in a weaker position than US firms because of trade limits and sanctions.
A wake-up call for America. Even if DeepSeekโs AI fizzles out, it should sound the alarm from Silicon Valley to Washington that US geopolitics is no longer up to task. Sanctions have largely failed, and the only way to protect American corporations from the next phase of Chinese exports may be to ban them. And while this protects US companies at home, what about around the world? At the same time, Chinese AI now matches American AI. What happened to US technological leadership?
Potential friction with Chinaโs strategic relationships. There is a growing possibility that DeepSeekโs AI models will drive a wedge between China and several other nations. This does not just have to be over censorship. India may not want Chinese AI in its society because of data collection. Russia may feel uncomfortable as it points to pulling Moscow deeper into a Chinese orbit. Instead of bolstering connectivity, AI may reverse it, accelerating the โvertical world.โ
Consumers matter as much as countries. Consumers' desire to have cheaper and more advanced AI systems at their fingertips could overpower their fears of national and economic security, similar to what is happening with TikTok. And this is exactly what China may be banking on. A new dynamic is forming, where consumers collide with their countries to keep accessing Chinese services, which governments believe are trojan horses.
CONCLUSION
DeepSeekโs AI is not just buzz or a fad.
Those who view it this way will remain oblivious to how profound and significant what is happening is. It makes sense why some are referring to it as a โSputnik Moment.โ
The status quo between the US and China has changed. And the timing is suspicious. It comes just after the US sought to limit global access to advanced AI chips, a shot at China. And it comes as the clock continues to countdown on TikTok, putting Chinese technology access to the US in the spotlight.
Now, however, the world is no longer watching China; it is watching the US. What will America do? What does this mean for US leadership? What else does China have up its sleeve?
While there are many aspects to DeepSeek's geopolitics, like censorship or market disruption, one that has garnered little attention is data. What is about to begin is a new fight to control data, as the West frets that China is training AI systems on the back of public data in Western societies (i.e. social media, encyclopedias, etc.). In the US outlook, if China is building advanced AI systems using American data, that is a non-starter.
DeepSeek has disrupted the US, the West, and the world. But in the process, it has ignited several fights with no clear end date. What can be gleaned, however, is that as the US and China fight over AI, it will end up dividing and separating the globe even further.
The geopolitics of DeepSeek is not just about a Chinese AI company or open-source AI systems. It is not about AI parameters or GPUs. It is a story about new innovation, geopolitical strategy, and, ultimately, how nations will function very differently in a world with more than one superpower.
ABISHUR/MR. GEOPOLITICS
Want to republish this insight? Letโs talk: mrgeopolitics@substack.com.
If you like Mr. Geopolitics, your colleagues might too.
Feel like doing something unique for somebody important?
Terms/Conditions
Mr. Geopolitics is a commercial service providing actionable insights to leaders. All Mr. Geopolitics content is the property of Abishur Prakash/The Geopolitical Business, Inc., and is protected under Canadian Copyright Law. This includes, but is not limited to: ideas, perspectives, expressions, concepts, etc. Any use of the insights, including sharing in the public domain, partly or wholly, requires explicit written permission. If you would like to discuss sharing Mr. Geopolitics with your audience, please contact: mrgeopolitics@substack.com.
"Geopolitical Conditionsโ is an expression/concept of Abishur Prakash/The Geopolitical Business, Inc.