Processing capability is becoming increasingly strategic to the success of advanced cloud services such as artificial intelligence (machine learning and deep learning) and quantum computing.
Cloud-based services are the growth engine for Technology giants AWS (Amazon), GCP (Google) and Azure (Microsoft). Similarly, Apple’s new content offerings (Apple TV+, Apple News+) and Apple Health are built on the cloud. Google and Microsoft continue to push more services to the cloud, the edge of the network and in the case of Google on-device (read our related articles about Google HERE and Microsoft HERE). Broadly-defined Artificial Intelligence – AI, machine learning, deep learning, Natural Language Processing – offerings are optimized with purpose-built hardware. Some examples follow.
One such example is Google Stadia – Google’s cloud gaming service – which has purpose-built chips designed by AMD. Another example is Google’s quantum computing effort (based out of Google Santa Barbara) which we wrote about HERE.
Google’s Tensor Processing Units (TPUs) are custom-developed integrated circuits that are optimized for machine learning/ deep learning environments that power Google services such as Search, Street View, Google Photos and Google Translate among others. TPU is a key component of Google’s data center strategy (the company will have invested approximately $13 billion in data centers in 2019).
AWS announced its “Braket” managed service which will enable customers to experiment with Quantum computing environments delivered as a cloud service.
Microsoft recently announced that Graphcore’s AI accelerator chips were incorporated into Azure. Microsoft invested in Graphcore’s $200 million round last year.