With GPUs, NVIDIA has been one of the biggest beneficiaries of the AI trend, but as tech giants have developed their own AI chips, including Google's TPU, Apple's neural engine, Microsoft's FPGA, and Amazon's ongoing customization for Alexa AI chip, NVIDIA can keep its leading position? It can maintain the current alarming growth, or will lose a large number of customers?
The rise of artificial intelligence (AI) has sparked a variety of arms race, with the largest technology companies seeking to benefit from emerging technologies. NVIDIA has been one of the biggest beneficiaries of this trend as its graphics processor GPU) is an early choice for training AI systems.GPU can perform a large number of complex math operations at the same time, making it the perfect choice for AI applications.
NVIDIA's data center business has seen explosive growth, achieving a three-digit year-over-year increase for seven consecutive quarters and its share price has risen by more than 1,000% since early 2015. Some well-known companies that once embraced NVIDIA GPUs are now seeking to develop other solutions Program to enhance or replace the GPU. Amazon may be the latest to enter the competition of technology giants.
Alexa's connection with the cloud
Amazon was the early adopter of artificial intelligence and, according to recent reports, Amazon is researching custom AI chips that can be processed on the device or processed at the edge, rather than just relying on the device to connect to the cloud.Nowadays, When the digital assistant Alexa makes a request, the information is transmitted to the cloud, which processes the request and sends the response back to the device.This process results in a slight delay.The ability to handle speech recognition locally will improve any device driven by the digital assistant, including Echo series smart speakers) response time.
Amazon boosted its processor capabilities by spending $ 350 million on Annapurna Labs, an Israeli chip maker, in early 2015. The company's network chips for data centers are capable of transmitting larger amounts of data while consuming less power Amazon currently has more than 450 employees with a degree of chip experience that the company may develop in other specialty chips, and the report also hinted that Amazon may be developing AI processors for its AWS cloud computing unit.
Google, Apple, Microsoft are developing programs to replace the GPU
Google revealed in early 2016 that it is developing custom AI chips known as tensor processors (TPUs), which are designed to provide more efficient performance for the company's deep learning AI applications that enable Learning by processing massive amounts of data, the chip laid the groundwork for TensorFlow, the framework used to train the company's AI system.
The latest version of the TPU handles both the AI training and reasoning phases, which, as its name suggests, "learn" during the training phase, and the reasoning phase is where the algorithms do their job of being trained.Google recently announced that customers of Google Cloud can now Access these processors.
Apple has long been a supporter of user privacy and has taken a different path from its technology counterparts, with the company's mobile device adding electronic noise to any data transmitted to the cloud while stripping out any personally identifiable information, thereby Provide a greater degree of user privacy and security.With the release of iPhone 8 and iPhone X, Apple has developed a neural engine, as part of its new A11 bionic chip, a chip that can handle a variety of AI Advanced processors that greatly reduce the amount of user information that is transmitted to the cloud and help to protect the data.
Microsoft earlier betting on a customizable processor - called a field programmable gate array (FPGA) - is a dedicated chip that can be configured for a customer's specific use after it has been manufactured, and these have become Microsoft Azure Cloud Computing System foundation and offers more flexible architecture and lower power consumption than traditional products such as GPUs.
NVIDIA growth continues
Although these companies have adopted different processor strategies, they are still using a large number of NVIDIA GPUs.
NVIDIA's growth continues. In the most recent quarter, NVIDIA posted a record revenue of 2.91 billion U.S. dollars, up 34% from a year earlier. The company's data center unit, which includes sales of AI, grew 105% %, Reaching 606 million U.S. dollars, currently accounting for 21% of NVIDIA's total revenue.
Competition is inevitable, but so far no solution has completely replaced the GPU. NVIDIA can now sit back and relax.