AI chip

The AI chip industry is booming. New technological developments, complex datasets, increasing adoption of deep learning algorithms and neural networks have accelerated the progress of AI chips. Big technology companies are producing new improvements in AI chip technologies. Here are some updates from top AI chip companies and recent developments.

Google is using AI to design chips faster than humans

Recently, Google has claimed that it is using AI technology to design chips faster than humans. The new technology could design the chips within six hours, whereas the same chip would take months for humans to develop. The company claims that it has already used the technology to develop the next iteration of Google's tensor processing unit chips which is used to perform AI-related tasks. Google's new AI technology can also draw the chip's floor plan. Normally, it would take months for humans to develop the chips, but the company's new technology can instantly plot the placements of different components like the CPUs and GPUs on the micro boards.

Mythic launches the industry's first Analog AI chip

The company recently launched its new M1076 Analog Matrix Processor (Mythic AMP), a single chip analog computation device w can deliver interesting compute resources at very low power. The AMP chip can pump out up to 35 TOPS (trillions of operations per second), with a low profile of 3W. The company is currently focusing on edge AI deployments where large or multiple models work simultaneously; hence, performance and power are critical. This innovation has proved to be a breakthrough for the AI chip industry.

SK Telecom's AI chip about to be tested for use at NHN servers

South Korean cloud serviced company is testing Sapeon, an artificial intelligence technology-based chip; developed by SK Telecom, to encourage the potential adoption of homegrown chips in the computer industry. This decision towards encouraging the adoption of domestically produced chips arose because of the uncertainty in the global supply chains.

NVIDIA GPU's continue to dominate Artificial Intelligence-based chips

Currently, graphics and gaming are a major part of NVIDIA's revenue. NVIDIA's GPUs were deployed back in 2019 and it has dominated 97.4% of the AI accelerators; for example, the hardware used to amplify the speed of data processing; at the top cloud service providers like AWS, Google, Alibaba, and Azure. Currently, 70% of the world's top supercomputers are using this GPU technology. NVIDIA is conscious about its approach in the AI industry, and its recent developments have proven that it is one of the forerunner companies.

Tenstorrent challenges NVIDIA to beat its best GPU technology

Chip start-up company Tenstorrent claims to build a GPU that can even beat the best GPU technologies provided by NVIDIA. Tenstorrent's processor is capable of 368 trillion operations per second (TOPS), beating the market leader NVIDIA's chip which serves up to 260 trillion. The chief goal of the company is to attract outside programmers who are attracted to groundbreaking AI applications for their software.