YouTube, faced with the limitations of Moore’s law, decided to create its own video chip called the Argos Video Coding Units (VCUs). The off-the-shelf chips they were using were not efficient for transcoding tasks. The company realized the need for an ASIC, a specialized chip designed for specific tasks. YouTube’s first video chip project was approved after a meeting with Susan Wojcicki. The Argos VCUs provided a significant performance boost compared to traditional server hardware. Google’s development of custom chips, such as the Tensor Processing Unit (TPU), also gives them a strategic advantage in the field of AI.
| Signal | Change | 10y horizon | Driving force |
|---|---|---|---|
| YouTube creates its own video chip | Shift from using off-the-shelf chips to custom ASICs | Increased video transcoding efficiency | Cost savings and strategic advantage |
| Google deploys Argos VCUs with significant performance boost | Improvement in transcoding performance | Enhanced video compression for better experience | Control over product roadmap of chip companies |
| Big tech companies building custom chips for strategic advantage | Custom chips offer specialized tasks and control over roadmap | Increased control and competitive advantage | Consolidation in chip industry and strategic advantage |
| Google’s development of Tensor Processing Unit (TPU) | Custom ASICs for AI applications | Advancement in AI technology and competition | Competitive advantage in AI field |
| Google’s use of custom-built Tensor SoC in Pixel 6 series | Integration of hardware and software | Seamless user experience and innovation | Optimization and differentiation |