DeepSeek users are finding the hot new generative artificial intelligence chatbot as good as its advertisements claim and it may provide an opportunity for the Middle East to accelerate its digital transformation, industry observers say.
The platform, which has grabbed worldwide headlines this week, has an open-source, free-to-use model for regular users and charges low fees for developers, and its large language model was trained using significantly less financing compared to billions spent by rival industry leaders on creating similar platforms.
It has received rave reviews for its performance, which has been reported to be better at particular functions compared to Anthropic's Claude and industry leader OpenAI's ChatGPT.
Its model now offers the potential to make companies reconsider their AI strategies, especially in the Middle East, where organisations are actively looking for cost-effective, localised and compliant AI solutions, said Sid Bhatia, a vice president and general manager for the Middle East, Turkey and Africa at New York-based AI company Dataiku.
"If DeepSeek delivers strong performance and multilingual capabilities – including Arabic – it could offer a compelling alternative to global AI players," he told The National.
DeepSeek’s cost-effectiveness and open-source platform can evoke "optimism" in the Gulf, as it would allow smaller companies that lack huge funding to build AI models, said Sophia Matveeva, chief executive and founder of London-based consultancy Tech for Non-Techies.
"Companies that wanted to invest in AI but couldn’t because they thought it was too expensive, will now get in the game," she told The National. "This means that digital transformation in the GCC will accelerate.
"For consumers, all the AI services you were paying premium prices for are about to get way less expensive and you are going to have more choice."
DeepSeek burst on to the scene over the past week as a cost-effective chatbot that, according to tests, outguns industry faces such as ChatGPT, Claude and Meta Platforms' Llama.
Its cost has also become a key selling point: its full offering is free to use, while the fee for its advanced platform for developers is only a fraction of what others charge.
The platform, made in China, also rattled stock markets, most notably AI chip bellwether Nvidia, wiping off more than $1 trillion from market caps and the wealth of tech bosses tied to AI, over fears of a stiff challenge to US leadership of technology. Tech stocks, however, have since recovered.
"The global AI ecosystem is taking note of DeepSeek’s developments ... innovation and the driving down of costs are key to unlocking AI and enabling mass adoption longer-term," said Richard Clode, a portfolio manager at asset management firm Janus Henderson.
DeepSeek is also able to "combine the best" of ChatGPT and Anthropic’s Claude, [as its] writing and reasoning style are smooth and sophisticated like that of Claude, and doesn’t have that robotic AI feel that ChatGPT does", Ms Matveeva said.
Her company has subscriptions to both ChatGPT and Claude, "because one [ChatGPT] is a good writer, the other [Claude] one scans the internet".
But with DeepSeek, "I only need one, which is way more convenient and a lot cheaper".
Kellie Blyth, a Middle East-based technology and privacy lawyer, and partner at law firm Addleshaw Goddard, called DeepSeek "very good", delivering similar results to GPT's newer reasoning models. s She said it is a "good substitute for other leading large language models"
She told The National: "The most impressive thing is the ability to generate an equivalent quality of results using a different approach and the willingness to make it available as an open source, so that others can use and deploy it."
However, companies are still weighing up the viability of DeepSeek, assessing its maturity, scalability and ecosystem integration before making major shifts in their AI investment, Mr Bhatia said.
Established models such as ChatGPT, Claude and Google's Gemini have proven reliability with potential advancements, but new entrants like DeepSeek can introduce fresh perspectives and advancements.
Each model is recommended to be assessed for long-term maintainability, scalability and infrastructure adaptability, "ensuring it can integrate seamlessly into evolving AI ecosystems", said Javier Alvarez, senior managing director at FTI Consulting.
"Security, privacy and regulatory compliance remain central considerations, alongside potential vendor lock-in risks and geopolitical factors that could impact long-term viability," he told The National.
