Homegrown artificial intelligence startup Krutrim introduced Graphics Processing Unit as a Service (GPUaaS) for enterprises and developers to train AI systems in early May.
GPUs are specialized computer chips that perform high-speed mathematical calculations and render graphics and images needed for AI processing. The demand for GPUaaS is expected to increase significantly as startups offer Generative Artificial Intelligence (GenAI) products that are the core software of his AI.
Click here to follow our WhatsApp channel
GPUaaS allows businesses to use expensive chips as needed and pay only for what they use. It is a flexible solution that leverages GPUs and is economical as businesses avoid investing in hardware.
“The vision is to enable developers to do cutting-edge innovation. The platforms and tools that developers need are compute and services, and that's what we've been opening up. For example, We have launched GPU-as-a-service, model-as-a-service and location services for people,” said Ravi Jain, head of strategy at Krutrim, which means “artificial” in Sanskrit. I am.
GPUaaS is not new, but it has gained traction in the last few years with the GenAI boom. “…(a) GPUaaS platforms accelerate tasks such as deep learning and rendering and streamline GPU resource management by automating allocation, scaling, and maintenance processes. Manually managing these tasks is It can be complex and time-consuming, especially at scale,” said Rashid Khan, chief product officer and co-founder of Yellow.ai, a customer service automation platform.
According to a Nasscom report, India had 18 GenAI startups in 2021, which more than doubled to 60 as of Q1 2023, leading to the largest share of the country's GPU-enabled computing infrastructure. Demand is emerging.
According to data from Nasscom, which represents India's technology industry, $8 billion was invested in AI between 2013 and 2022 across 1,900 AI startups in India ($3.2 billion in 2022 alone).
According to Allied Market Research, the international GPUaaS market was worth $1.83 billion in 2021 and is projected to expand at a compound annual growth rate of 33.2% to reach $19.56 billion by 2030.
Indian demand
Experts say that GPUaaS will be useful for Indian companies as they have just started implementing GenAI and are cautious about investing in computing power.
“Most enterprises use AI-as-a-service from service providers and are not exposed to the cost of GPUs. A few people who want to deploy their own custom AI services are “This is where AI infrastructure as a service is positioned,” said Naresh Singh, senior analyst at market research firm Gartner. Masu.
Experts also said that cloud service providers can offer GPUaaS to enterprises, startups, global competency centers and governments in India.
“Instead of investing in in-house GPU capacity, which is expensive and requires specialized skills, enterprises are better off hiring GPUs on demand from these cloud providers,” said S Anjani Kumar, consulting partner at Deloitte India. It makes sense.”
Small organizations can use GenAi through GPUaaS. Ramprakash Ramamoorthy, his director of AI research at ManageEngine, an information technology (IT) service, said: He says, “GPUaaS providers are making GPUs more accessible by leveraging their bulk purchasing power to negotiate better terms, supporting a wide range of enterprises in leveraging advanced computational technologies.” company.
As part of the Rs 10,000-crore India AI Mission, the government may borrow GPUs from domestic companies and provide the chips to startups at subsidized rates, according to media reports.
Indian data services company Yotta Infrastructure is offering GPU servers at competitive prices of $2 to $2.5 per hour of usage to help global giants such as Amazon and social media platform owner Meta Platforms. We aim to compete with. Worldwide, these costs range from $9 to $12.
“The government is going to procure these GPUs as a service model from people like us who are building data centers by deploying GPUs and putting a software layer on top,” Yotta's top executive said. Director Sunil Gupta said earlier.
According to experts, governments consider GPUaaS to be a good fit for their strategic needs. “Government customers typically have sovereign needs and strict security compliance requirements. Typically, they may start with AI-as-a-Service, but from a regulatory, national security, and strategic perspective. They can also take the strategic step of investing in their own infrastructure, which could drive market demand for AI infrastructure (captive or as a service),” said Gartner Senior Director Analyst. said one Naresh Singh.
Pay and leave
The main benefit of GPUaaS for governments is its “pay-as-you-go” capability. This eliminates the need for upfront investments in hardware, software, skills and services, and allows governments to pay as they are used, he said. Deloitte India.
Data center and cloud companies are expanding their GPUaaS businesses as demand increases. Cloud computing company CtrlS plans to invest $2 billion in AI and data center development over the next six years.
“Our expansion plans include the integration of advanced technologies such as liquid cooling, AI, heating, ventilation, and air conditioning (HVAC), and optimized rack layouts to meet the growing demand for AI applications and cloud services. ” said Ranjit Metrani, president of the company. CtrlS Data Center Managed Services.
Neysa, an AI cloud company focused on delivering GenAI solutions as a service, raised $20 million in seed funding in April. The service will help clients “discover, plan, deploy and manage their GenAI projects cost-effectively in a consumption-based model,” the company said earlier.
Technology-focused venture capital funds believe that government support that gives them access to setting up GPU clusters and training AI models at subsidized prices will help startups and academia.
service report
> GPUaaS allows organizations to “pay only for what they use” for specialty chips
> Global GPUaaS market expected to be worth $19.56 billion by 20301
> Businesses using GPUaaS save on infrastructure costs
> 10,000 rupees Indian AI mission allows GPU rental in the country
> Provide subsidized chips to startups
> Data and cloud companies expand GPUaaS business
