Connor Leonard, Rachel Ranjith & Zara Pribicevic
Artificial intelligence has evolved into more than a mere scientific advancement; it has transformed into a competitive arena for efficiency and worldwide dominance. Central to this evolution is ChatGPT, previously the preeminent leader in conversational AI, now confronted by a significant rival: DeepSeek. With its efficiency-driven approach and open-source architecture, DeepSeek heralds a shift in AI development, raising questions about the balance between adaptability and specialisation.
However, the ramifications extend beyond artificial intelligence models. NVIDIA, the leader in AI chip manufacturing, has experienced a challenge to its supremacy as DeepSeek demonstrates the availability of alternatives. Meanwhile, as the U.S.-China AI rivalry increases, innovation has turned into a geopolitical hotspot. With dynamic changes in the AI realm, how does DeepSeek contest ChatGPT, and what influence are felt by NVIDIA? How is the overarching AI weapons race changing global power dynamics?
The ChatGPT Phenomenon
A true marvel of modern invention, ChatGPT, the golden child of OpenAI’s GPT architecture, has become the standard for conversational AI. Alongside its ability to generate human-like text, answer complex questions and assist with tasks, ChatGPT’s ability to provide personal and professional responses have exponentially multiplied its user base, ranging from students to suburban moms to 9-5 workers. We can trace the success of this model back to its scale; trained on vast amounts of data, it uses a generalist approach to handle a wide array of tasks. Like any true everyday hero, it can write your college essay, debug your code or help you draft a breakup text.
However, the unfortunate downfall of this ‘everything-tool’ lies in its defining generalist approach. While ChatGPT excels at breadth, it, more often than not, struggles with depth, especially in specialised domains. Its reliance on pre-2021 data for GPT-3 also limits usability. These limitations, along with increasingly high subscription prices, highlight a growing tension in AI development. Will ChatGPT manage to stay relevant while the pendulum of user demand swings back and forth between versatility and expertise?
DeepSeek: A New Challenger
Cheaper and better products from China never come as a surprise, and yet, the advent of DeepSeek seemed to shock U.S. AI oligarchs. Capable of ChatGPT-esque speech (since they share the same foundational technology) while also offering a broader spectrum of knowledge, DeepSeek distinguishes itself through a more focused approach. DeepSeek is designed to prioritise efficiency and precision, often excelling in cases where ChatGPT’s generalism might fall short. However, once again, the strengths of this model are also its downfall, as DeepSeek’s narrower focus comes with trade-offs. By optimising specific tasks, it may lack the versatility that has made ChatGPT so popular.
DeepSeek’s emergence suggests a growing recognition that one-size-fits-all solutions may not be sufficient as AI applications become more diverse and complex. However, this shift toward specialisation also risks creating a fragmented AI ecosystem, where users must navigate multiple tools for different tasks, rather than relying on a single, unified platform. The challenge for DeepSeek, then, is to prove that its targeted approach can deliver enough value to justify the added complexity.
Differences, Implications and Competitive Advantages
A key asset of DeepSeek’s groundbreaking reveal was its computational efficiency. While ChatGPT and other US AI oligarchs invest billions in AI chips for model training, DeepSeek achieves almost superior performance using only a quarter of the AI chips required for ChatGPT. This efficiency causes a ripple effect in multiple spheres. It significantly reduces cost of operations. For an increasingly climate conscious consumer, this efficiency results in a lower resource footprint that could serve as a major competitive advantage. Additionally, the reduced requirement for expensive state-of-the-art AI chips significantly lowers the barrier to entry, making it easier for smaller businesses to access the necessary hardware for deploying advanced AI models.
Another key differentiator is DeepSeek’s open-source nature. This is contrasted by OpenAI’s proprietary model which, ironically enough, started off open-source (lending to the brand name, OpenAI). DeepSeek’s openness of a superior model strips some competitive advantage from ChatGPT, as it enables broader access to creating models of similar intelligence. It must be stated, however, that an open-source model is not without fault. Without centralised oversight, there may be significant challenges in maintaining consistency, preventing misuse, quality control, security, and ensuring ethical standards.
The implications of DeepSeek are understandably widespread and have both erased and created competitive advantages on both sides of the competition. The impact can also be seen in industries adjacent to AI, such as NVIDIA, the lead supplier of coveted AI chips.
NVIDIA: An Overview
NVIDIA has been one of the most remarkable success stories in the stock market over the past few years. The company, originally known for manufacturing high-performance graphics processing units (GPUs) for gaming, has transformed itself into an artificial intelligence (AI) powerhouse. This pivot has driven NVIDIA’s stock price to unprecedented levels, with its market capitalisation surpassing $1 trillion in 2023, making it one of the most valuable tech companies in the world.
A key driver behind this explosive growth has been the surge in demand for AI-related computing power. NVIDIA’s cutting-edge chips, particularly its H100 and A100 GPUs, are essential for training and deploying large AI models. This dominance in AI infrastructure has led to exponential revenue growth, allowing NVIDIA to post record-breaking earnings and maintain its stronghold in the semiconductor industry.
AI has become inseparable from advanced semiconductor technology. Training large-scale AI models, such as OpenAI’s GPT-4 and Google’s Gemini, requires vast computational resources. NVIDIA’s GPUs, with their parallel processing capabilities, are the gold standard for AI workloads, significantly outperforming traditional central processing units (CPUs).
The reliance on NVIDIA’s technology for AI development has made the company indispensable to major tech firms. However, the company is also reliant on AI firms continuing to use their chips. An analyst from UBS estimated that Microsoft, Meta, Amazon, and Alphabet collectively make up 40% of NVIDIA revenue.
NVIDIA’s Stock Crash: The Facts
Recently, NVIDIA’s stock experienced a sharp decline amid concerns over the emergence of DeepSeek, a new AI player backed by Chinese tech firms. DeepSeek was able to develop a GPT with a combination of NVIDIA chips and cheaper alternatives. The use of cheaper alternatives was previously thought to be impossible in the short term, thus the demand for NVIDIA chips is uncertain in the future.
The stock dropped 17% to $188.42 in one day, erasing nearly $600 billion of value – the largest one day drop in U.S. history. It briefly recovered up to $128.99 over the next few days, but recently it seems to have found an equilibrium.
The future value of the company is still uncertain. The quality of copycat AIs is uncertain, as well as what effect, if any, they will have on NVIDIA’s biggest buyers. Microsoft, Meta, and Alphabet have already built their software around NVIDIA chips which makes a decoupling seem unlikely. However, there are also broader geopolitical aspects to this issue.
DeepSeek, AI, and the Great Geopolitical Chessboard
Let us rewind to how AI rose to power to become a tool in political debate. The first AI Safety Summit was organised by then-UK Prime Minister Rishi Sunak, and the outcome was stellar. The Bletchley Declaration was this summit’s achievement – 28 countries, including the United States and China, signed the agreement, promising international co-operation in order to manage the challenges and risks that AI posed.
The tables have now turned, and at the most recent AI Summit in Paris a new declaration was put forward: “A pledge for Trustworthy AI in the World of Work”, reiterating the importance of transparency and promoting AI accessibility in order to avoid a digital divide. 60 countries signed the declaration, including China, however the UK, who was once an advocate for AI safety, followed suit with their special partner the US, who also refrained from signing the agreement. Vice President JD Vance emphasised that he had concerns over “authoritarian censorship”. The Paris AI summit has shown a change in the dynamics, with an emphasis on geopolitical rivalry and the never ending pursuit of economic and technological advantage.
The United States: Not Amused
Let’s talk about efficiency. This January, while President Trump made Elon Musk’s schedule busier than ever with the introduction of the Department of Government Efficiency, a Chinese firm used creative methods to increase artificial intelligence efficiency – DeepSeek. Across the pond in Washington, DeepSeek’s success has raised some serious eyebrows. The US has been cranking up the pressure on China’s AI ambitions, raising tariffs and slamming the door on high-tech chips, warning allies against playing too nice with Beijing.
Trusted advisor J.P. Morgan and their Chairman of Market and Investment Strategy Michael Cembalest noted at a talk on the Global Market Outlook that although Deep Seek is a compelling story, it may be too good to be true. Cembalest emphasised that it was difficult to tell what really happened with DeepSeek, as Chinese statistics are not to be trusted. He compared them with his middle school child’s report, insinuating that we do not have the full picture of DeepSeek’s success story. We have to wonder though, is the US trying to play it down, and are they possibly too nonchalant for their own good?
China: Playing the Long Game
For Beijing, DeepSeek’s rise to fame isn’t just a win – it’s a strategic flex. Following years of U.S. sanctions blocking access to high-end microchips, China found itself at a crossroads: play nice and roll over or get creative. It chose the latter, giving the country a tech boost. The long game for China begins with DeepSeek, but The People’s Republic is aiming high for full scale AI-driven industrial innovation, aligning with the new policy “Made in China 2025”, emphasising China can not only scale and commercialise emerging technologies, but also innovate and be at the forefront of the field. Beijing isn’t necessarily aiming for the crown in the AI chatbot arena, but to employ the underlying technology to create cost-effective, commercially feasible solutions, which it can subsequently export to lower-income nations. China’s objective is not exclusively “frontier AI”, but rather “mass-market AI”.
Europe: Stuck in the Middle
Meanwhile, Europe is still trying to decide what to do with its seat at the table. On one hand, leaders like former UK Prime Minister Rishi Sunak organised the first ever AI Safety Summit in November 2023. French President Emmanuel Macron is advocating for AI advancements leveraging the region’s nuclear energy edge, and the most recent “AI Investment Plan” pledging around €200bn towards accelerating AI development. Conversely, Europe seeks to avoid entanglement in the conflict between two technological superpowers. The European Union, hindered by internal discord and slow innovation processes, risks becoming a mere observer, rather than a player, in the race for AI, which is now a key economic driver.
Microchips: The Tiny Giants Shaping Global Power
At the core of this AI turmoil lies the true arbiter: microchips. The United States has adopted a stringent approach by prohibiting the shipment of advanced semiconductors to China, aiming to impede Beijing’s progress in artificial intelligence. However, China is not lamenting; it is ingeniously managing its resources to maximise the performance of the chips it can still utilise. When the United States intensified its restrictions, China retaliated by restricting the export of gallium and germanium, essential ingredients for semiconductor manufacturing. What is the outcome? A semiconductor rivalry in which both parties are striving to surpass one another.
The Bigger Picture: A Tech Cold War in the Making?
DeepSeek’s emergence is but a fragment of a larger geopolitical conundrum. The AI arms race transcends the mere development of superior models; it encompasses the control of future technologies, economic dominance, and global impact. The United States and China are in conflict, Europe is striving to maintain its significance, and the global community is preparing for the enduring ramifications of a fragmented AI ecosystem. Will this be the dawn of a new digital Iron Curtain? Will innovation discover a method to dismantle obstacles? AI is not only transforming sectors; it is also redefining global power dynamics.
Most of these developments point to a future where AI is both an economic driver and a geopolitical chess piece. In many ways, this entire situation is a microcosm of a broader global transition. ChatGPT’s success and subsequent dethroning proves the need for the precarious balance between being an all-encompassing model and trying to be everything to everyone. DeepSeek’s efficiency-led approach and open-source philosophy promises new paths forward but also exposes fresh challenges. NVIDIA’s stock woes highlight how dependent entire industries are on AI’s progress — yet also how quickly that progress can redefine winners and losers. Nations grapple for control over microchips, data, and AI capabilities, while global alliances and trade deals strain under the weight of competing ambitions. As technology becomes a barometer of power, the potential for fragmentation — specialised models, trade restrictions, or strategic self-reliance — grows. What is certain is that AI is actively reshaping the balance of power — politically, economically, and technologically — well beyond chatbots and chip manufacturers.