Self-sufficiency is "crucial"! Microsoft invests heavily to strengthen its self-developed AI models
Microsoft plans to expand its physical infrastructure to train self-developed artificial intelligence (AI) models, aiming to compete with companies like OpenAI and Anthropic. Mustafa Suleyman, Microsoft's head of consumer AI, stated that self-sufficiency is crucial for the company and revealed that large-scale investments will be made to enhance computing power. Microsoft will also deepen its collaboration with OpenAI while working with other model manufacturers. Microsoft CEO Satya Nadella emphasized the adoption of a multi-model strategy to meet customer needs
According to Zhitong Finance APP, Microsoft (MSFT.US) plans to expand its physical infrastructure to train its own artificial intelligence (AI) models, hoping these models can compete with companies like OpenAI and Anthropic. Mustafa Suleyman, Microsoft's head of consumer AI, stated during an all-hands meeting on Thursday that the company will make "massive investments" in its computing power clusters to train models.
According to attendees, Suleyman told employees that it is crucial for a company of Microsoft's scale to have the capability to be self-sufficient in the field of artificial intelligence. He added that Microsoft has adopted a strategy of deepening its collaboration with OpenAI, working with other model manufacturers, and building its own models.
The world's largest software manufacturer primarily relies on OpenAI's large language models to support its products. However, as both companies launch competing products and seek more partners, signs of tension have emerged in their relationship. Suleyman, a co-founder of DeepMind, joined Microsoft last year to lead the development of the company's self-developed models and created a consumer franchise project focused on artificial intelligence.
Last month, Microsoft released its first large language model developed under Suleyman's leadership. The model was trained on 15,000 NVIDIA H100 chips. Suleyman added that the computing clusters used by cutting-edge models developed by companies like Meta Platforms Inc., Alphabet Inc.'s Google, and Elon Musk's xAI are 6 to 10 times larger than Microsoft's, indicating that Microsoft's model creation efficiency is higher.
Microsoft CEO Satya Nadella added that the company's plan is to adopt a multi-model strategy for all its products, selecting the appropriate AI models based on customer preferences. Earlier this week, media reported that the company plans to use Anthropic models in some of its products.
Additionally, it was reported that Microsoft and OpenAI announced on Thursday that they have signed a non-binding agreement on new terms for their relationship, which will allow OpenAI to advance its restructuring plan to become a for-profit company
