What's Next in AI: Our vision for the future of AI hardware
Since 2012 the computation requirements for large AI training jobs have grown more than 300,000x. To proliferate progress, we must overcome long-standing bottlenecks in scientific discovery. CPU advancements alone are no longer the primary source of competitive advantage in the world of large-scale, complex computing. In coming years, more specialized accelerators beyond the GPU will enable significant performance advantages.
At IBM Research, we’re developing a new class of AI hardware that is inherently energy-efficient, increasing compute power by orders of magnitude without the demand for increased energy. By leveraging the power of AI and the flexibility of the hybrid cloud, we can comprehend and apply massive and growing bodies of scientific knowledge.
Since 2012 the computation requirements for large AI training jobs have grown more than 300,000x. To proliferate progress, we must overcome long-standing bottlenecks in scientific discovery. CPU advancements alone are no longer the primary source of competitive advantage in the world of large-scale, complex computing. In coming years, more specialized accelerators beyond the GPU will enable significant performance advantages.
At IBM Research, we’re developing a new class of AI hardware that is inherently energy-efficient, increasing compute power by orders of magnitude without the demand for increased energy. By leveraging the power of AI and the flexibility of the hybrid cloud, we can comprehend and apply massive and growing bodies of scientific knowledge.
