South Korean AI chipmaker FuriosaAI has rejected Meta’s $800M acquisition offer, choosing to remain independent to focus on its cutting-edge AI chip development.

Oracle is also behind the biggest AI training project out there - Stargate

Artificial Intelligence (AI) has experienced unprecedented expansion in recent years. The spread of AI tools has exceeded the boundaries of data centers and high-performance devices to enter embedded systems ranging from IoT sensors to portable medical devices, up to the most complex industrial applications. Running AI models efficiently in embedded environments has some limitations due to the scarce computational and energy resources available.

How to put the pieces together in a complex design with AI is an unsolved problem.

The hyperscale cloud has evolved from racks of commodity x86 servers into one of the most heterogeneous compute environments ever built. As enterprises pursue increasingly demanding workloads—such as AI training, 5G packet processing, and large-scale data analytics—the classical CPU-centric architecture is no longer sufficient. Cloud providers now rely on an ecosystem of accelerators—GPUs, FPGAs, DPUs, and confidential computing hardware—to unlock performance, efficiency, and new levels of security.

Taichi has demonstrated high efficiency and complex AI task handling, using a novel hybrid optical approach to overcome scaling issues.

The new IBM z17 mainframe features AI capabilities across hardware, software, and systems operations.