CPUs are an affordable choice, and those working on AI models and those working on AI models, despite the hype surrounding CPUs (GPUs), ubiquitous central processing units (CPUs) are far from being discontinued in the age of artificial intelligence.
A CPU, sometimes called the “brain” of a computer, is the main processor of any computer, and the GPU is a chipset that has become popular to support AI models. At the beginning of March, the Ministry of Electronics was talking about subsidizing GPU access to a rate of 1 pound per hour, significantly less than 2 pounds to 2.40 pounds per hour when launching a new general computing facility. This raised the question of whether the CPU will be completely replaced by a GPU in the future.
Ramesh Loganathan, head of the Centre for Innovation and Entrepreneurship at IIIT Hyderabad, noted that the rate is granted only to certain providers.
“The different providers are better than what you get in the market.
Similarly, Vice President Tech of Corover.ai, who owns BharaTgpt, and Vikas Tripathi, the product, pointed out that CPUs can cost between 50,000 and 1 pounds on a good machine, while GPUs can cost between 25-50 Lakh depending on use case.
Tripathi was also solid about CPU persistence on Tech Space, which even today states that it still needs a CPU.
Use case specific
“The CPU is always an integral part of a personal computing machine or laptop, so you cannot replace a GPU with a GPU.
Tripathi said Corover uses a wide range of GPUs for Bharatgpt, but uses CPUs for highly domain-specific in-house SLMs.
NTT Data, a global IT and business services company that spoke about launching GPUs as service around October 2025He said there are still a lot of workloads that require CPU. The company had expected future workloads, a hybrid of two chipsets. It uses current usage of the CPU around the Networking equipment facility. At a recent media conference in Mumbai, he said that while GPUs will play a key role in the future, the CPU is likely to stick out in the long run.
