Thomson Reuters Develops Proprietary AI With $40M Investment
Thomson Reuters' shift to proprietary AI mirrors IBM’s cloud strategy, both aiming for increased technological sovereignty.
What Changed
Thomson Reuters has announced a $40 million investment to develop its own language model, "Thomson," using Alibaba's Qwen as a foundation. This marks the first instance where Thomson Reuters has opted to develop an internal AI solution rather than depending on external providers such as OpenAI or Anthropic. Historically, IT companies have rented AI capabilities to enhance operations, but the current move highlights a strategic pivot towards owning proprietary technology.
Strategic Implications
By investing in a proprietary AI model, Thomson Reuters could increase its competitive edge specifically in legal and financial sectors. The new model is optimized for its content repositories, such as Westlaw, potentially outperforming generalized models. This initiative also signals a decrease in reliance on large AI firms, like OpenAI, thereby enhancing internal control over capabilities.
What Happens Next
Should the "Thomson" AI model succeed, expect other enterprises to follow suit, investing in tailored AI technologies that better serve their unique datasets. Within the next 24 months, it is likely we will see policy shifts aimed at encouraging domestic AI development, spurred by competitive advantages like those Thomson Reuters aims to achieve. Major tech firms may intensify efforts to offer modular AI optimizable by smaller players to retain relevance.
Second-Order Effects
The move could encourage similar strategies among competitors, impacting cloud-based service models that provide rented AI capabilities. Alibaba's position as a tech supplier may strengthen if more corporations seek to customize their AI models upon its frameworks. Meanwhile, digital content stakeholders might face new regulatory landscapes prioritizing domestic AI growth.
Free Daily Briefing
Top AI intelligence stories delivered each morning.