Foundation models are the driving force behind the generative AI revolution. For most enterprises, now and in the coming years, acquiring foundation models from external providers will be the norm, rather than training their own from scratch. These models serve as critical elements in the larger AI framework, yet it’s important to recognize that not every foundation model is of the same caliber. In our previous global report, we introduced an evaluation framework on the capabilities that technology architecture and delivery practitioners should focus on, spanning technology, business, and ecosystem. Among these capabilities, model adaptation towards specific business needs and local availability in the region are especially important for firms in APAC.

Our initial research into foundation models within the APAC market highlights both parallel developments and distinct paths when compared to the West across key sectors. The unique strengths identified in APAC are molding the region’s potential future directions in the field. Tech leaders and AI practitioners aiming to develop generative AI capabilities in the APAC region must acknowledge the unique landscape in which they operate. The diversity of customer demands and regulatory requirements varies significantly from country to country within APAC. Additionally, the region’s vast linguistic diversity necessitates a specialized emphasis on multilingual-understanding capabilities. It’s also crucial for these leaders to recognize that, although global AI giants such as Anthropic, Google, and OpenAI have led the international AI race, APAC is home to a robust selection of foundation models. Notably, Chinese vendors have established a dominant presence in this space. Meanwhile, the development of foundation models in Japan, India, and Korea is accelerating quickly, contributing to the region’s competitive landscape in AI.

The foundation model ecosystem in APAC is characterized by three distinct groups of vendors: first, hyperscalers, which utilize their extensive cloud infrastructures to advance the development of foundation models; second, AI platform providers, which place a strong emphasis on integrating foundation model support into their product development strategies; and lastly, foundation model startups, which adopt open-source approaches to discover and capitalize on market opportunities. To get more detail on the key trends of foundation models in APAC or to discover representative vendors in each vendor group, Forrester clients can download the report or request an inquiry with Charlie Dai and Leslie Joseph. I will also kick off the landscape report of AI foundation model for languages (aka large language models) for the APAC market in May following our global version — please stay tuned!