In the world of artificial intelligence, the rise of DeepSeek has captured everyone’s attention, stirring up quite the buzz among major AI entities, including Google. In a recent meeting, Demis Hassabis, CEO of DeepMind, addressed the concerns among his team about this emerging Chinese AI startup. His message? No need to panic, as Google’s dominant position in AI technology remains unchallenged, according to CNBC reports.
Hassabis made these remarks during an all-hands meeting in Paris. Alphabet CEO Sundar Pichai read out a question raised by one of the employees. The query highlighted the rapid success of DeepSeek and sought insights on what Google might learn from their breakthrough.
To give some background, DeepSeek rolled out its open-source model powered by R1 V3, causing quite a stir as NVIDIA’s market shares took a significant hit, plummeting by $600 billion in just one day. A research document suggests that DeepSeek’s model outshines OpenAI’s o1 reasoning model in numerous areas like science, mathematics, and coding. Even more intriguing is how they achieved these results at a fraction of the typical development cost for proprietary AI models.
Despite this splash, Hassabis was quick to temper the narrative, labeling DeepSeek’s success as somewhat “overstated” after a detailed examination. He speculated that the cost of model training touted by DeepSeek is likely just “a small portion” of the actual expenses, hinting that the startup utilized more hardware than initially disclosed.
This discussion comes on the heels of reports claiming DeepSeek invested $1.6 billion and acquired 50,000 NVIDIA GPUs for its AI model development.
In the words of DeepMind’s Hassabis: “We’re confident in our more efficient, high-performing models compared to DeepSeek’s. Our strategy is solid, ensuring we uphold our leadership position in AI through this year and beyond.”
While Hassabis acknowledged Google’s substantial lead over DeepSeek in AI, he admitted the Chinese team is one to watch. Nonetheless, he pointed out potential obstacles for DeepSeek, such as security issues and geopolitical factors, that could impede their progress and prevent them from realizing their full potential.
For further context, DeepSeek recently experienced a significant cyberattack, hinting at some of the security challenges faced by the startup.
Now, it’s worth considering if Google might have missed a golden opportunity in the AI race. As generative AI continues to evolve rapidly, the question of who will emerge as the frontrunner remains unanswered. Roman Yampolskiy, an AI safety researcher, has an intriguing viewpoint: “If you have enough money to buy sufficient computing power, you could develop AGI today.”
While Yampolskiy’s perspective adds an interesting dimension, Microsoft’s CEO Satya Nadella doesn’t see Google leading the AI charge. Last year, he remarked that although Google had all the potential to dominate AI with its vast resources, they ultimately fell short. He acknowledged Google’s impressive capabilities, saying they possess everything from data to silicon, enabling them as a vertically integrated player.
In response, Sundar Pichai had a sharp retort aimed at Microsoft’s collaboration with OpenAI. He said, “I’d love to see a side-by-side of Microsoft’s models and ours any day. They’re using external models.”
This comment refers to the partnership between Microsoft and OpenAI, where Microsoft gains access to cutting-edge AI models in exchange for providing computing power and financial backing.
Interestingly, Microsoft’s Nadella described DeepSeek’s AI development as “remarkably impressive” and suggested that the progress coming out of China should be taken seriously by everyone in the AI field.
