- AiNews.com
- Posts
- SoftBank Eyes TSMC for AI Processors as Intel Partnership Falters
SoftBank Eyes TSMC for AI Processors as Intel Partnership Falters
Image Source: ChatGPT
SoftBank Eyes TSMC for AI Processors as Intel Partnership Falters
SoftBank’s ambitious Project Izanagi, which aims to develop AI processors, has encountered challenges with its current partner, Intel. According to sources cited by the Financial Times, Intel has struggled to meet the ‘volume and speed’ requirements set by SoftBank. As a result, the Japanese conglomerate is now considering a partnership with TSMC, the world’s largest contract chipmaker. While discussions are ongoing, no agreement has been finalized, partly due to TSMC’s existing commitments to major clients like AMD and Nvidia.
Project Izanagi’s Ambitious Goals
Project Izanagi is part of SoftBank’s broader strategy to challenge Nvidia’s dominance in the AI GPU market. The initiative includes plans to build AI processors, develop a proprietary software stack, and construct massive AI data centers powered by dedicated energy sources. Despite Intel’s inability to meet SoftBank’s needs, insiders suggest that the door remains open for future collaboration.
Challenges in Chip Design
One challenge for SoftBank is its lack of in-house chip design expertise. While it owns Arm, which develops ‘Compute Subsystems’ (CSS) reference designs, the company has not previously designed chips at the scale required for Project Izanagi. SoftBank also acquired Graphcore, a hardware company known for its competitive AI processors, but Nvidia’s superior CUDA software stack has kept Graphcore from gaining significant traction. Intel had offered its chip design services to SoftBank, though it remains unclear whether this would have been an Arm-based design or if it could be completed within the necessary timeframe.
Masayoshi Son’s Broader AI Vision
Masayoshi Son, SoftBank’s CEO, is determined to push forward with his AI vision, which includes positioning the company as a direct competitor to tech giants like AWS, Google, Microsoft Azure, and Meta. This move could drastically alter the competitive landscape in the AI industry. Financially, the project could cost tens to hundreds of billions of dollars, and Son has reportedly approached investors in Saudi Arabia and the United Arab Emirates for potential funding, though no concrete deals have been made.
Potential Risks and Strategic Partnerships
In an effort to secure support, Son has also been in discussions with major tech companies like Google and Meta. He hopes to attract backing to counter Nvidia’s dominance in the AI processor sector. However, this strategy could put SoftBank in direct competition with these companies, which might complicate potential partnerships.
A significant risk in Son’s plan is the potential impact on Arm’s relationships with its key clients. Critics argue that if Arm moves into direct chip design and production, it could strain its relationships with developers of Arm-based data center processors. Nevertheless, SoftBank appears willing to take this risk in pursuit of its ambitious goals.