Chineese AI Revolution Sparks Chaos
The release of Chineese A.I. model Deepseek R1 has wiped a Trillion dollars of value off U.S. Companies in less than a week
In Auckland, New Zealand we had a three day weekend, and honestly I’ve just been glued to my phone the whole time, watching global turmoil unfold in the tech scene and financial markets caused by the release of “Deepseek R1”, a Chinese A.I. to rival Chat GPT.
It was only a few days ago that I sent out my last “Rumination”, about the release of Operator, the intelligent agent from Open AI, however now I feel compelled to provide another blog post on A.I - such is the speed at which things are moving.
For those confused why the US state and Silicon Valley are having a meltdown:
China released multiple Al models that are 50x more efficient than the best American Al models and made them open source, ruining the American “for profit” Al market
The new model, DeepSeek-R1, achieves comparable performance to top chatbots at a fraction of the cost, challenging assumptions about AI development resources.
Key Points:
DeepSeek-R1 matches or exceeds leading AI models in reasoning and mathematical tasks
DeepSeek's open-source approach allows researchers to freely use and modify their models
This development has caused concern among investors in major US tech companies
"With the release of R1, DeepSeek essentially cracked one of the holy grails of Al: getting models to reason step-by-step without relying on massive supervised datasets. Their DeepSeek-R1-Zero experiment showed something remarkable: using pure reinforcement learning with carefully crafted reward functions, they managed to get A.I. models to develop sophisticated reasoning capabilities completely autonomously. This wasn't just about solving problems— the model organically learned to generate long chains of thought, self-verify its work, and allocate more computation time to harder problems
The technical breakthrough here was their novel approach to reward modeling. Rather than using complex neural reward models that can lead to"reward hacking" (where the model finds bogus ways to boost their rewards that don't actually lead to better real-world model performance), they developed a clever rule-based system that combines accuracy rewards (verifying final answers) with format rewards (encouraging structured thinking). This simpler approach turned out to be more robust and scalable than the process-based reward models that others have tried."
Deepseek erases $1.2Trillion of US stocks
Over a trillion dollars in market value has been wiped off U.S. companies in the last 7 days, as the ripple effects of the implications of Deepseek’s technology cascade through the tech sector.
Industry giants like Nvidia, who have long been at the forefront of providing hardware solutions for AI development, have seen their share prices plummet.
Nvidia experienced a significant dip as investors scramble to assess the implications of a technology that dramatically reduces the cost of AI operations, perhaps by a hundred fold, and reduce the need for the most expensive A.I. chips. Deepseek R1 is capable of running on a desktop computer.
The rapid devaluation of tech stocks has left analysts questioning whether this marks the beginning of a broader market correction or a pivotal shift in the AI landscape.
Project Stargate
Project Stargate was announced by President Donald Trump on January 21, 2025. This expensive $500 Billion initiative is a collaboration between OpenAI, led by CEO Sam Altman, SoftBank Group, and Oracle Corporation. The project’s primary goal is to invest in artificial intelligence (AI) infrastructure across the United States.
Project startage now looks ridiculous when Deep seek R1 was trained for just $5.5Million USD. China has released something Open source at 1% the cost of America’s “pay for it” model.
It truly shows what insane times we are living in when the US can announce a $500 billion dollar investment in A.I, and in less than a week that looks outdated. To say that things are moving at light speed would be an understatement. And let’s not forget that once A.I. is as good at improving itself as humans are, A.I. may simply improve itself virtually instantly to a level that we can’t understand with human level intelligence.
What Makes Deepseek Different?
Deepseek has captured global attention for several reasons:
Open Source Accessibility: Unlike many proprietary AI models, Deepseek has been released as open-source software. This move has democratized access to cutting-edge AI technology, allowing developers, researchers, and businesses to integrate and build upon its capabilities without the heavy licensing fees often associated with such tools.
Unprecedented Cost Efficiency: Deepseek is remarkably cheaper to run than its closest competitor, ChatGPT. Early reports suggest that the cost of operating Deepseek is only a fraction of what similar models require, making high-powered AI applications more accessible to smaller organizations and startups.
Lower Training Costs: One of the most astonishing aspects of Deepseek is how much less expensive it was to train compared to other large language models. Leveraging optimized architectures and innovative training methods, Deepseek’s creators achieved a level of performance that rivals the best in the industry, at a fraction of the cost. This efficiency has triggered a reevaluation of the financial assumptions underpinning the AI sector. By being extremely close to the hardware and by layering together a handful of distinct, very clever optimizations, DeepSeek was able to train these incredible models using GPUs in a dramatically more efficient way. By some measurements, over ~45x more efficiently than other leading-edge models. DeepSeek claims that the complete cost to train DeepSeek-V3 was just over $5m. That is absolutely nothing by the standards of OpenAl, Anthropic, etc., which were well into the $100m+ level for training costs for a single model as early as 2024.
Why This Matters
Deepseek’s release signals more than just a new player in the AI space; it represents a fundamental shift in how AI systems are developed, deployed, and monetized. Here’s why it’s such big news:
Eroding Competitive Advantages: Companies like OpenAI, which have relied on proprietary models and premium pricing strategies, now face the challenge of competing with a model that is not only cheaper but also openly accessible and free to download.
Leveling the Playing Field: By lowering costs and eliminating access barriers, Deepseek empowers a broader range of entities to leverage AI. This democratization could lead to an explosion of innovation, as smaller players enter a field previously dominated by tech giants.
Disrupting the Hardware Market: Deepseek’s efficiency threatens the dominance of hardware providers like Nvidia, whose high-performance GPUs have been a cornerstone of AI development. If more efficient models reduce the demand for expensive hardware, the entire ecosystem could shift. I have already seen demos of Deepseek running locally on an iPhone 16
Deepseek is still the People’s republic of China
Out of interest, I asked Deepseek about Tiananmen Square, and no surprises, it wouldn’t give me an answer. Interestingly, it did give me an answer, which flashed on the screen for a second, and was then removed.
Also, bear in mind that when you deal with US companies (Facebook, etc, they are using and stealing all your data (if you’re not paying for something, you’re not the customer, you’re the product being sold). And the same is true for Chinese companies (Tik Tok, Deepseek etc).
The beautiful thing about Deepseek R1 is that being open source, I assume that I can download it, and then remove any Chinese filtering if I have the skills. And if not, Deepseek has shown coders the way to build their own AI’s without filtering in the future.
What Comes Next?
The release of Deepseek has undoubtedly reshaped the AI landscape, but the full implications are yet to unfold. While some celebrate its democratizing potential, others warn of the destabilizing effects it could have on established players. Regulators, investors, and industry leaders are now grappling with the question of how to adapt to this new reality.
One thing is certain: the AI revolution is far from over, and the chaos sparked by Deepseek may only be the beginning. As innovation accelerates and market dynamics shift, the next few months will be critical in determining the future trajectory of artificial intelligence.
You made the gist of complex subject easy to follow. I like that you make key points summarising.