sb.scorecardresearch

Union Budget 2025 in association with

logo4 logo1 logo2 logo3 logo5 logo6

Published 20:31 IST, January 28th 2025

Could DeepSeek's Open-Source Language Model Help Smartphone Industry?

The smaller variants of DeepSeek's open-source large language model (LLM) will be more advantageous for smartphones, as AI has gained no traction with consumers so far, Jefferies, the investment banking and capital markets firm highlighted in its latest report.

Follow: Google News Icon
  • share
DeepSeek
DeepSeek is a Chinese startup that has disrupted the AI industry with its AI assistant. | Image: Shutterstock

The smaller variants of DeepSeek's open-source large language model (LLM) will be more advantageous for smartphones, as AI has gained no traction with consumers so far, Jefferies, the investment banking and capital markets firm highlighted in its latest report. The report, however, added that DeepSeek's success refers to another company that has made strides in AI, possibly with a more capable model or larger data set. 

DeepSeek's progress is seen as a hopeful sign for the future of AI, suggesting that it may be possible to overcome the limitations of smaller models like Apple's.

"If smaller models can work well, it is potentially positive for smartphone. We are bearish on AI smartphone as AI has gained no traction with consumers. More hardware upgrade (adv pkg+fast DRAM) is needed to run bigger models on the phone, which will raise costs. AAPL's model is in fact based on MoE, but 3bn data parameters are still too small to make the services useful to consumers. Hence DeepSeek's success offers some hope but there is no impact on AI smartphone's near-term outlook," Jefferies added. 

The report pointed out the return on investment (RoI) in AI as DeepSeek's LLM is considered as an efficient model that could reshape the future of AI development.

DeepSeek is a project created by the successful AI-driven quant fund High-Flyer and has developed a large language model (LLM) that rivals the performance of industry giants like GPT-4 in less computational cost. 

Highlighting the RoI concern, the report added, “The market naturally will worry about demand growth in computing power. We have been highlighting our concern about AI's RoI, as the massive investment in GPUs (eg, just NVDA's 2024 GPU rev could be $200bn) has generated little return.”

DeepSeek's model is trained at a cost of just $5.6 million which is a stark contrast to expenses associated with other top-tier models. 

The report highlighted that the huge investments made so far in the AI technologies have not given expected returns, as there is still a lack of tangible AI monetization to justify such enormous outlays.

Talking about two possibilities on AI investments, the report added, “We believe DS's success could drive two possible industry strategies: 1) still pursue more computing power to drive even faster model improvements, and 2) refocus on efficiency and ROI, meaning lower demand for computing power as of 2026.” It added that as investors look more closely at the bottom line. AI companies especially those in the US may find themselves under increased pressure to justify their rising capital expenditures, the report added.

Updated 20:31 IST, January 28th 2025