Deepseek has aroused AI ecosystem under the leadership of the US with its latest model, shaving hundreds of billions in the leader of Chip Nvidia Market cap. While the sector leaders are falling out, fewer AI companies see the opportunity to increase with Chinese startups.
Several companies associated with AI said to CNBC that Deepseek’s appearance is a “massive” opportunity for them, not a threat.
“The developers are very eager to replace the open and closed models of open code such as Deepseek R1 …,” said Andrew Feldman, chip chip chip chip chip chip chip chip chip chip Systems.
The company competes with NVIDIA graphic treatment and offers cloud-based services through its own computer clusters. Feldman said that the publication of the R1 was created by one of the greatest spikes of Cerebras who ever sought for their services.
“R1 shows it [AI market] Growth will not dominate one company-Moats hardware and software do not exist for open code models, “Feldman added.
The open source refers to the software in which the original code is freely available on the web for possible changes and redistribution. Deepseeek models are an open source, unlike competitors such as Openi.
Deepseek also claims that its R1 reasoning model subdires the best American technologies, despite lower costs and is trained without top graphic units for processing, although they are observers and competitors in industry and competitors in the industry questioned these claims.
“As in PC and Internet markets, the price drop helps to encourage global adoption. AI market is on a similar path to secular growth,” Feldman said.
Chips concluding
Deepseek could increase the adoption of new chip technology by accelerating the AI cycle from training to “Conclusion phase”, said start-up chips and experts in industry.
The conclusion refers to the act of using and applying AI to make prediction or decisions based on new information, not the construction or training of the model.
“Simply put, AI training refers to the construction of tools or algorithm, while the conclusion is actually about using this tool to use in actual applications,” said Phelix Lee, analyst from capital in Morningstar, with an emphasis on semiconductors.
While Nvidia has a dominant position in the GPUs used for AI training, many contestants see spreading room In the “conclusion” segment, where they promise greater efficiency for lower costs.
AI training is very intense, but the conclusion can work with less powerful chips that are programmed to perform a narrow range of tasks, Lee added.
Numerous AI chip startups have told CNBC that they have seen more demand for the inquisition of chips and computing while clients adopt and build on the Deepseek Open Code model.
“[DeepSeek] He has shown that smaller open models can be able to be able to be able or more capable of larger models, and this can be done with a fraction of costs, “said Sid Sheth, CEO of AI Chip Starting-up D-Matrix.
“With a wide availability of small capable models, they catalyzed the age of concluding,” he told CNBC, adding that the company recently saw the increase in the interests of global customers who want to accelerate their conclusion plans.
Robert Wachen, co -founder and Coo from AI Chipmaker Etched, said dozens of companies have reached the startup since Deepseek has announced his reasoning models.
“Companies are 1738900134 Moving their consumption from cluster clusters to cloning clusters, “he said.
“Deepseek-R1 has proven that calculating the conclusion is now [state-of-the-art] The approach to any large model and thinking supplier is not cheap – we will only need more and more calculating capacity to scaling these models for millions of users. “
Jevon’s paradox
The analysts and experts in the industry agree that the achievements of Deepseeka amplifiers for AI conclusion and the wider AI chip industry.
“The deepseek performance seems to be based on a series of engineering innovations that significantly reduce the conclusion costs, at the same time improving training costs,” he states report from Bain & Company.
“In a bull’s scenario, the current improvement of efficiency would lead to cheaper conclusion, encouraging greater acceptance of AI,” he added.
This pattern explains Jevon’s paradox, a theory in which the reduction of costs in the new technological urge increased demand.
The financial services and investment company Wedbush said last week in a research note that he still expects to use AI throughout companies and retail consumers globally to encourage demand.
Speaking CNBC -ov “quick money” Last week, Sunny Madra, Coo from Groq, developing a AI conclusion chips, suggested that, as the overall demand for AI is growing, smaller players will have more room for growth.
“How will the world need more tokens [a unit of data that an AI model processes] Nvidia cannot give everyone enough chips, so it gives us the opportunity to sell even more aggressively on the market, “Madra said.
