How Apple could work with DeepSeek to pull ahead in the AI race

Ever since DeepSeek burst onto the scene last month, there’s been no shortage of opinions about what the Chinese startup’s artificial intelligence accomplishments mean for America’s AI giants like OpenAI, Microsoft, Google, and Meta. 

However, there’s one company that’s usually been absent from any discussion of just how bad DeepSeek’s arrival is for many of America’s tech giants: Apple. Whenever a tech insider or analyst mentions Apple and DeepSeek together, its usually to suggest that the arrival of the Chinese LLM could be beneficial to the iPhone maker. 

Is that really the case? AI technology is moving so quickly (DeepSeek virtually appeared out of nowhere) that it seems futile to make long-term predictions about any advancement’s ultimate impact on the industry, let alone an individual company. Still, I can see a few ways that Apple may benefit from DeepSeek and its successes.

Partnering with DeepSeek can give Apple Intelligence a way into China

The most direct way that Apple could benefit from DeepSeek’s arrival is if the company decided to actually partner with the Chinese startup.

It’s no secret that Apple badly wants to bring Apple Intelligence to China, as the company’s software chief, Craig Federinghi, told me last June when it first previewed its AI platform. China is Apple’s second-largest market after the US. 

Apple believes that the addition of Apple Intelligence to its phones represents a unique selling point that will attract new iPhone users and encourage existing ones to upgrade, leading to increased sales. Increasing sales in China is very important to Apple, where its phones have been struggling recently against domestic rivals (the Cupertino company recently saw its China sales drop 11.1% on an annual basis).

The problem for Apple is that China has stringent laws surrounding AI systems—even those designed for consumers. That’s why, to meet the country’s legal requirements, Apple might likely partner with an already approved AI provider in China. Indeed, in December, Reuters reported that Apple was in early talks with Tencent and ByteDance about the possibility of integrating their AI models into Apple Intelligence in the country.

But now Apple may have a third option: DeepSeek. From both a technological and optical perspective, choosing the small startup’s LLM for integration with Apple Intelligence in the country could be a wise move.

Technologically, integrating DeepSeek into Apple Intelligence on iPhones in China could confer several performance benefits. Two significant advantages of DeepSeek over other LLMs are that it can run on less powerful chips and uses less energy. This makes it perfect for Apple’s Apple Intelligence platform, which tries to perform as much on-device computing as possible to make things speedier and more private for its users.

Partnering with DeepSeek over other AI providers in China may also be excellent optics for Apple from a marketing standpoint, at least locally. Ever since DeepSeek upstaged the much better-funded U.S. tech giants, the small startup has become a point of pride in the nation. Its founder, Liang Wenfeng, even reportedly received a hero’s welcome when he returned to his hometown of Zhanjiang over the Lunar New Year holiday.

If Apple wants to increase sales in China, what better way than by touting that the AI developed by China’s tech hero is powering Apple Intelligence?

AI R&D just got cheaper for Apple

Even if Apple doesn’t partner directly with DeepSeek, the Chinese startup’s disruption of the U.S. AI industry seems to offer some indirect advantages for Apple.

Before DeepSeek’s arrival, general wisdom held that if a company wanted to be competitive in the AI race—to be in it to win it—it needed to invest tens of billions of dollars in research and development. After all, that’s what OpenAI, Microsoft, Meta, and Google have done. Apple could have thrown the same amount of money into AI development in order to catch up with its U.S. rivals—but DeepSeek’s success shows that the company may not have to do that anymore. 

DeepSeek says that its model was trained for less than $6 million, a pittance compared to the money OpenAI and Meta, for example, have spent training their models. And DeepSeek’s R1 LLM outperforms OpenAI’s and Meta’s models in many areas.

If it’s now known that powerful LLMs that can compete with the heavyweights can be created and trained for just cents on the dollar, Apple no longer needs to consider spending tens of billions to catch up and can direct its massive cash stockpiles elsewhere. 

Further, DeepSeek claims it took just two months to train its model. If that’s the case, then Apple will be able to catch up to its U.S. competitors in no time.

Apple’s position in the AI race

So, does DeepSeek’s arrival give Apple a leg up in the AI race? 

When it comes to potentially bringing Apple Intelligence to China and boosting iPhone sales in the country, it very well may. But outside of China, it’s impossible to say if DeepSeek’s arrival will actually materially benefit Apple’s position in the AI race long-term. 

DeepSeek’s innovations seem to show that you don’t need to spend billions of dollars and years training advanced AI models. This revelation doesn’t only benefit Apple, however. While Apple now might not have to spend the time and money its US competitors did to create a model on the scale of ChatGPT and Llama, neither do other companies now, which levels the playing field for everyone, big and small.

Besides, who knows what other AI breakthroughs are just around the corner? DeepSeek virtually came out of nowhere. There’s no guarantee it won’t be upstaged by another groundbreaking AI advancement in the months ahead that disrupts the industry all over again. And because of that, it’s impossible to say who will be in what place on the future AI leaderboards—no matter if you’re talking about Apple, OpenAI, or DeepSeek itself.

https://www.fastcompany.com/91273725/deepseek-benefit-apple-intelligence-china-ai-race-openai-google-microsoft-meta?partner=rss&utm_source=rss&utm_medium=feed&utm_campaign=rss+fastcompany&utm_content=rss