Necessity drives innovation, and when resources are limited, creativity takes over. China’s DeepSeek exemplifies this with its latest R1 open-source AI reasoning model, a breakthrough that claims to deliver performance on par with U.S.-backed models like Chat GPT at a fraction of the cost. By defying conventional wisdom, DeepSeek has shaken the industry, triggering a sharp selloff in AI-related stocks. Yet, disruption breeds opportunity. While established players scramble to adapt, this shift could accelerate AI accessibility, leveling the playing field for new innovators.
Let’s start with the technology itself. What did DeepSeek accomplish?
DeepSeek’s approach used novel ways to slash the data processing requirements needed for training AI models by leveraging techniques such as Mixture of Experts, or MoE. When given a problem to solve, the model utilizes a specialized sub-model, or expert, to search for the answer rather than utilizing the entire model. This has allowed DeepSeek to create smaller and more efficient AI models that are faster and use less energy.
This innovation impacts all participants in the AI arms race, disrupting key players from chip giants like Nvidia to AI leaders such as OpenAI and its ChatGPT. It also impacts power providers like Vistra and hyperscalers—Microsoft, Google, Amazon, and Meta—that currently dominate the industry. Investors fear DeepSeek’s advancements could slash demand for high-performance chips, reduce energy consumption projections, and jeopardize the huge capital investments—totaling hundreds of billions of dollars—already poured into AI model development.
Still, some industry players view the DeepSeek announcement as an opportunity rather than a threat. Semiconductor giant ASML CFO Roger Dassen summed up the news at a press conference following release of Q4 2024 financial results when he said “If you make it cheaper, and if you make it more accessible, if you make it more energy efficient, it can lead to a democratization of AI in a way that is necessary in order for to this to become prevalent.”
However, questions remain over DeepSeek’s methodologies for training its models, particularly concerning the specifics of chip usage, the actual cost of model development (DeepSeek claims to have trained R1 for less than $6 million), and the sources of its model outputs.
The uncertainty surrounding DeepSeek’s model training methods is a key concern among AI experts. Some suspect that DeepSeek used advanced U.S. model outputs to refine its models through distillation.
AI Czar David Sacks believes DeepSeek may have stolen intellectual property from the U.S. “There’s substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI’s models,” Sacks said in an interview on Fox News. Distillation is a machine learning technique that transfers knowledge from a large model to a smaller model. If the accusations are confirmed, the result will likely be additional sanctions on the exports of U.S. technology to China.
Regardless of the ethics and possible repercussions, DeepSeek’s advancements will only accelerate the growth and adoption of AI —not curtail it. By significantly reducing the costs associated with model development, DeepSeek’s techniques will ultimately make AI more accessible to businesses of all sizes. This democratization of AI technology could promote innovation and application across various industries.
Steve Cohen, founder of Point 72 Asset Management, believes the long-term repercussions are positive for the AI industry. “Our view is what happened with DeepSeek is actually bullish because it advances the move to artificial super-intelligence,” Cohen said Tuesday at the iConnections Global Alts conference in Miami Beach.
For investors, the pressing question is whether the AI giants—Microsoft, Google, Amazon, and Meta—can justify the return on their existing AI investments. One potential threat is that the high-margin oligopoly profits among AI beneficiaries could decline as competition increases. Massive capital expenditures may no longer serve as an effective barrier to entry if model development costs plummet, which is one potential outcome from the DeepSeek news.
Despite the new threat, Microsoft is fully committed to its AI strategy. “We are innovating across our tech stack and helping customers unlock the full ROI of AI to capture the massive opportunity ahead,” said Satya Nadella, chairman and chief executive officer of Microsoft, in a press release on FY25 Q2 earnings. “As AI becomes more efficient and accessible, we will see exponentially more demand,” Nadella noted.
Yes, DeepSeek’s breakthrough introduces uncertainty for industry leaders, but it also has the potential to accelerate AI innovation at an unprecedented pace. As development costs decline, AI adoption will expand, fueling economic growth and technological advancements. While established players may face shrinking profit margins and increased competition, the broader economy stands to gain from enhanced productivity and efficiency. Stock prices may fluctuate in the short term, but the long-term impact of AI becoming more affordable and accessible will drive greater benefits—sooner and at a lower cost.
Read the full article here
