The rise of DeepSeek shook tech giants’ comfortable position as the leaders driving AI innovation. The Chinese start-up showed how model efficiency can be achieved with less money and fewer resources.
BeInCrypto spoke with ten industry leaders on why the technological sector had spillover effects on the crypto market and how DeepSeek’s rise has permanently redefined the future of AI development.
A Bucket of Ice Water for American Tech Giants
The rise of DeepSeek and its deep-set effects on the crypto market served as a wake-up call to Western technology sectors that they no longer have a far-reaching upper hand on AI development.
Only two weeks ago, the Chinese start-up released two AI models: R1 and V3. These systems proved as efficient as those developed by tech giants OpenAI and Google—even ranking higher in some metrics. They were also produced at a fraction of the cost.
While language-learning models (LLMs) like Meta’s Llama 3.1 cost upwards of $60 million to produce, DeepSeek earned itself headlines by reducing the cost of training a frontier model to just $6 million.
Just hours after DeepSeek’s launch, the news wiped a trillion dollars off the market capitalization of leading US technology firms. Nvidia, the world’s dominant supplier of AI chips, saw its value fall by $600 billion.
The US stock market suffered its worst single-day loss ever, and crypto felt the impact. DeepSeek’s arrival caused major declines in mining stocks like Marathon and Riot, which heavily rely on Nvidia hardware.
The news also triggered a $1 billion crypto sell-off, with Bitcoin dropping 5% and altcoins seeing even steeper 8-10% declines. Meanwhile, AI-driven cryptos saw a 10% drop in market capitalization over 24 hours, with four of the top five AI coins suffering heavy losses.
DeepSeek’s emergence humbled overconfident tech giants. It raised scrutiny over their overreliance on billion-dollar investments and future revenue growth.
It also showed that any future disruptions in the race for innovation will inevitably have a spillover effect on the crypto market.
China Creates DeepSeek With All Odds Against it
DeepSeek rocked the markets because it showed that China wasn’t all that behind the United States in the race toward the most efficient AI models. Until the news hit on January 27, the tech stocks for major players like Microsoft, Google, and OpenAI showed a positive sentiment.
This sensation was primarily based on the fact that these tech giants are established and well-funded. They already have a solid market position and access to the most refined hardware and software needed to drive AI innovation.
“These companies not only have a technological edge but also the infrastructure, vast datasets, and financial resources to maintain their dominance,” said Pavel Matveev, Co-founder of Wirex.
Meanwhile, during Joe Biden’s presidency, Nvidia was barred from selling its GPU processors to China. These export restrictions forced China to rely on the stockpile it had built up until that point.
Despite these challenges, China created DeepSeek.
“Due to US export restrictions, the Chinese didn’t have anywhere near the access to the hardware that US companies did. But again, this is economics 101: scarcity of resources leads to innovation, or “needs must,” for the rest of us. China had to go down a super deep level of engineering and truly innovate. It’s really a triumphant story,” said Sebastian Pfeiffer, Managing Director of Impossible Cloud Network.
For Yang Tang, CEO of QStarLabs, something like this was bound to happen.
“This is a natural evolution in technology development: a scrappier competitor that used a better process to achieve better outcomes. To note, everything DeepSeek did was previously published in academic and/or industry research. This definitely will force established AI labs to think differently as many have been overly research-focused,” he said.
It also taught the Western world a valuable lesson.
Sometimes Less Really Is More
A year ago, OpenAI CEO Sam Altman predicted that the AI industry would require trillions of dollars in investment to fund the development of specialized chips. These chips are essential for powering the energy-intensive data centers that support the industry’s increasingly complex AI models.
Other leading technology companies recently took similar initiatives. Meta has already announced that it plans to spend as much as $65 billion this year to expand its AI infrastructure. The company aims to end the year with over 1.3 million graphics processors.
Microsoft announced plans for approximately $80 billion in data center development for fiscal 2025. Meanwhile, Amazon expects its projected 2025 spending on similar infrastructure to exceed its estimated $75 billion investment in 2024.
Many of these companies also stockpile GPUs and related AI hardware. Meta CEO Mark Zuckerberg, for example, said his company aimed to bring its GPU supply to 600,000 by the end of 2024.
Meanwhile, DeepSeek used a little over 2,000 Nvidia GPU units and $6 million to power its R1 model.
“DeepSeek’s breakthrough in reducing development costs and optimizing AI models with minimal computational resources signals a seismic shift in the competitive AI landscape. Traditional giants like Nvidia, OpenAI, and Google, which rely on massive computational power and expensive infrastructure (such as high-end GPUs and extensive cloud services), may find their traditional advantage in resource-heavy AI development diminishing,” Trevor Koverko, Co-founder of Sapien.io, told BeInCrypto.
Western companies’ realization that China was not too far behind in the race also spooked investors in traditional financial circles and crypto markets.
DeepSeek’s Impact on the Crypto Market Explained
The broader market downturn –especially in traditional markets– reflected a recalibration of expectations around tech valuations rather than a simple correction.
“The market had priced in aggressive growth assumptions for AI technologies, particularly around computational demands that would benefit companies like Nvidia and major cloud providers. DeepSeek’s breakthrough in achieving comparable results with less computing power has forced investors to reassess these assumptions,” said Karan Sirdesai, CEO and Co-Founder of Mira Network.
Though the crypto sector has no direct ties with DeepSeek, it does share a playing field with AI developers. As a result, crypto was just as impacted by the news of the R1 launch.
According to Sirdesai, the relationship between crypto and AI markets is more complex than simple correlation. While both fall under the technology umbrella, they operate on fundamentally different principles.
“Bitcoin and crypto valuations are rooted in monetary dynamics, network adoption, and regulatory landscapes, while AI developments center on technological capabilities and commercial applications,” he explained.
Nonetheless, crypto and AI both have a large presence in the technology sector.
“Both sectors compete for computational resources, especially GPUs, creating supply chain links. Plus, many investors are active in both spaces, so sentiment can spill over. When major tech companies see volatility from AI developments, it can ripple through to crypto markets through this shared investor base,” Sirdesai added.
The recent market movements following the release of DeepSeek’s R1 model attest to how susceptible the crypto market is to the overall sentiment of the technology sector.
“This interaction reflects a cultural and technological synergy between AI and crypto, suggesting that developments in one sphere can significantly influence the other,” added Forest Bai, Co-founder of Foresight Ventures.
As a result, closely following how American technology powerhouses respond to DeepSeek’s latest innovation will be crucial to understanding how similar events may impact the crypto market in the future.
A Period of Recalibration for American Tech Companies
The drop in investor confidence reveals uncertainty about the AI market’s future. These doubts center on whether computational scale will remain the key to competition and how efficiency innovations will reshape the sector.
“The AI race is no longer about who has the most GPUs but who can train the smartest, most efficient models. DeepSeek’s breakthrough proves that innovation in training can disrupt the AI monopoly,” Ilan Rakhmanov, Founder of ChainGPT, told BeInCrypto.
Rakhmanov highlighted the key technical innovations that DeepSeek implemented to side-step barriers to accessing GPUs.
“DeepSeek’s R1 model likely achieves its efficiency through a combination of optimized architecture, alternative training methods, specialized hardware, and energy-efficient compute strategies. By refining transformer efficiency, utilizing model sparsity, and incorporating retrieval-augmented generation, DeepSeek reduces computational demands without compromising performance. Its reliance on self-supervised learning, synthetic data augmentation, and reinforcement learning minimizes dependency on massive datasets, while custom AI accelerators or non-GPU alternatives help lower compute costs,” he explained.
To that point, Anthony Simonet, Head of Research at iExec, added:
“It employs techniques like its Mixture-of-Experts architectures, low-precision training, and knowledge distillation to maximize efficiency with fewer resources, enabling AI to run smoothly on standard hardware and making it more accessible,” he said.
Tech experts also quickly noted that DeepSeek published the research behind its model for the public to see.
The Case for Decentralized AI
In contrast to the traditional secrecy of US companies like OpenAI, DeepSeek impressively released its R1 model as completely open-source. Many industry leaders applauded this move, indicating that, for the future of AI to remain in the hands of the public, overall access must remain decentralized.
“DeepSeek has been a game-changer for the AI industry, and I believe it’s exactly the kind of wake-up call companies like OpenAI need. OpenAI was originally founded to make advanced AI accessible to everyone, but over time, we’ve seen a shift toward closed, gatekept models. The AI space is evolving, and DeepSeek has reminded us all of something important—great technology should be built for everyone, not just a select few,” said Rakhmanov.
Smaller developers with fewer resources welcomed this news. Access to DeepSeek’s design and research papers will allow them to refine their models without exhausting their research budgets.
“DeepSeek’s cheaper models reduce the GPUs required for training AI models, thus lowering computational costs. This efficiency allows AI to scale more affordably, making it accessible to businesses and researchers with limited resources,’ said Ron Bodkin, Co-founder of Theoriq.
With the start-up’s model now open-source, developers will analyze it extensively, driving further AI innovation.
“Since DeepSeek is open-source, the shift in the AI race will irrevocably shift more into the open-source arena, destroying the closed-sourced foundational model narrative. Being open-source benefits everyone, AI companies (every player in the ecosystem), innovators, and consumers. The only losers are the ones that cling to the closed-source model, which will see a rapid breakdown in the near future,” Steven Pu, Co-founder of Taraxa, told BeInCrypto.
As AI becomes less expensive and more accessible, it will become more of a commodity.
Commoditizing AI Technology
On the day of DeepSeek’s launch, Microsoft CEO Satya Nadella posted about Jevons’ Paradox on social media.
“Jevons paradox strikes again! As AI gets more efficient and accessible, we will see its use skyrocket, turning it into a commodity we just can’t get enough of,” Nadella said on X.
Also known as the rebound effect, Jevons’ Paradox is an economic principle coined by the English economist William Stanley Jevons. Increased efficiency in resource use can lead to increased consumption of that resource.
Applied to AI, as these systems become more efficient, the demand for their tasks may increase—a phenomenon that the increasing accessibility of AI research could amplify.
“Lowering costs, both in training as well as inference stages, is good. In technology, lowering costs have always led to wider adoption and higher overall consumption, not less. When cars became affordable, more people owned cars. When mainframes were shrunken into affordable personal computers, it drove the digital revolution. In just the same way, we’ll see more innovators and start-ups experiment with AI now that it’s become more affordable, leading to more utilization of AI, and higher demand for AI-related infrastructure such as GPU hardware,” Pu told BeInCrypto.
For Pfeiffer, the commodification of AI infrastructure will also change the nature of innovation tech companies will now seek. Developers once focused on creating the most refined LLM models. Now, efforts will shift to integrating this technology into various industries.
“DeepSeek trained on OpenAI and was able to build significantly on the progress of others. The LLM landscape will be commoditized and, most likely, fully open sourced. However, this is not where most innovation will happen. Indeed, the growth and evolution of AI will be seen less on the development side, but through the integration and use of AI. Vertical, deep integrations into industries and access to their data will matter much more than sophisticated LLMs because they are commoditized and their innovation progress will slow down,” he said.
This paradox could also give American tech giants an advantage over countries with restricted access to computational resources.
The US’s Upper Hand
Although DeepSeek’s most recent model has demonstrably narrowed the competitive gap between established US companies, the company is not immune to challenges.
According to Jevons’ Paradox, increased demand for AI products will also inevitably lead to increased demand for the resources needed to develop them. Though other alternatives are still being explored, GPUs will continue to be vital for the future development of AI technologies.
“DeepSeek also appears to be hitting capacity that limits their ability to scale their offering- they have limited sign-ups for their app to Chinese residents and their API is much slower than when they launched. I believe that they are unable to secure additional GPUs to allow them to scale their offering,” said Bodkin.
DeepSeek’s breakthrough also doesn’t dissolve the United States’ decades-long dedication to the development of AI infrastructure.
“Despite DeepSeek’s optimization breakthroughs, the AI race is still largely dictated by access to massive datasets, computational power, and end-to-end ecosystem control. Companies like OpenAI and Google don’t just rely on brute-force scaling—they also have proprietary data, cloud infrastructure, and extensive deployment pipelines. While alternative methodologies are promising, they will only disrupt the status quo if they can consistently outperform traditional approaches across diverse use cases. Right now, it’s too early to say whether DeepSeek represents an industry shift or simply an incremental improvement within an already competitive landscape,” Matveev told BeInCrypto.
Given this reality, Sirdesai believes the market reaction to DeepSeek was somewhat overblown.
“The market reaction seems to discount the complexity of commercializing AI technology. DeepSeek’s more efficient architecture is significant, but successful AI deployment requires robust infrastructure, strong security measures, and proven reliability in production environments. Western tech companies have spent years building these capabilities,” he said.
DeepSeek’s rise has undeniably reshaped the AI race, demonstrating that innovation can emerge from unexpected corners and challenge established giants.
As the industry continues to evolve, monitoring the interplay between open-source models, resource accessibility, and competition dynamics will undoubtedly shape the future of AI development and its impact on the world.
Disclaimer
Following the Trust Project guidelines, this feature article presents opinions and perspectives from industry experts or individuals. BeInCrypto is dedicated to transparent reporting, but the views expressed in this article do not necessarily reflect those of BeInCrypto or its staff. Readers should verify information independently and consult with a professional before making decisions based on this content. Please note that our Terms and Conditions, Privacy Policy, and Disclaimers have been updated.
