Microsoft Corp.’s artificial intelligence (AI) blueprint took shape on product and political fronts this week. Shortly after the software giant introduced new AI models that simultaneously process text, images and speech at a fraction of the computing power required by existing systems, the company urged the Trump administration to ease export restrictions on AI chips produced by U.S. companies.
Microsoft’s new Phi-4 models, announced Wednesday, deliver the performance of large language models (LLMs) twice their size, the company claimed in a technical report. Phi-4-Multimodal comes with just 5.6 billion parameters, and Phi-4-Mini with 3.8 billion parameters.
“These models are designed to empower developers with advanced AI capabilities,” Weizhu Chen, vice president of generative AI at Microsoft, wrote. “Phi-4-multimodal, with its ability to process speech, vision, and text simultaneously, opens new possibilities for creating innovative and context-aware applications.”
The new model’s technical chops could appeal greatly to enterprises that are increasingly seeking AI models that run on standard hardware or at the edge to slash costs and latency without compromising data privacy.
“By leveraging the Mixture of LoRAs, Phi-4-Multimodal extends multimodal capabilities while minimizing interference between modalities,” Microsoft said in the report. “This approach enables seamless integration and ensures consistent performance across tasks involving text, images, and speech/audio.”
In another blog post Thursday, Microsoft urged President Donald Trump to simplify export limits on chips in India, Israel, Switzerland, and elsewhere that can be used in data centers for training AI models. It argued the unintended consequence of the proposed system from former President Joe Biden would force those countries and others to turn to China for AI infrastructure.
In fact, China is using the proposed rule to convince U.S.-friendly countries that it would be a better long-term partner for AI infrastructure than the U.S., Microsoft President Brad Smith told the Wall Street Journal. “Their message is these countries can’t rely on the U.S., but China is willing to provide what they need,” he said. “That is not good for American business or American foreign policy.”
Chinese AI startup DeepSeek, which sent the tech industry and U.S. lawmakers into a tizzy with a low-cost, advanced AI model and is set to introduce an update soon, is one of several countries in China that Microsoft views as having strong potential, Smith added.
The Trump administration, which has successfully pressed tech giants Apple Inc., Meta Platforms Inc., and others to build AI data centers in the U.S. and invest heavily in research domestically, is reviewing the concerns of Microsoft, NVIDIA Corp. and industry groups that oppose Biden’s export restrictions.
