Artificial intelligence is creating a powerful new demand cycle for hardware, even as concerns grow around rising infrastructure costs, energy consumption and the economics of deploying AI at scale, Sriram Sitaraman, chief information officer, Synopsys, told ET.

Synopsys, a semiconductor design software company with $86 billion in market value, develops electronic design automation (EDA) tools and chip design intellectual property used by many of the world’s leading semiconductor companies.

According to Sitaraman, the rapid adoption of AI is driving demand across graphic processing units, central processing units, storage and data-centre infrastructure, creating a cycle between software and hardware.


“The use of AI is absolutely creating the market for AI hardware. GPUs, CPUs, storage, all of those are increasing,” he said.

While companies such as Nvidia, TSMC and SK hynix have emerged as some of the biggest beneficiaries of the AI boom, Sitaraman believes the opportunity extends beyond a handful of industry leaders.

“I don’t know who’s going to win the race, but there’s going to be a lot more hardware requirements that are going to pop up,” he noted, pointing to the wave of large-scale AI infrastructure investments being announced by companies such as OpenAI and Nvidia.

The surge in AI adoption is also putting pressure on power and cooling infrastructure, which has emerged as one of the industry’s biggest concerns. Sitaraman acknowledged that energy and cooling constraints are real but said they are unlikely to remain permanent bottlenecks.

“There are constraints today because the rate of change has not kept up with the rate of innovation in certain places,” he argued. “But I am sure that at some point there is going to be innovation in energy, innovation in cooling, that’s going to match the requirements.”

On India’s semiconductor push, he highlighted the role of government funding and workforce development in supporting the growth of the local ecosystem.

He pointed to government-backed semiconductor initiatives and industry skilling efforts as positive developments. “India has semiconductor funds. It’s been funding some local communities. We have established a lot of training centres to accelerate India’s semiconductor ecosystem and talent pipeline,” he said.

India has been one of Synopsys’ largest engineering and R&D bases for years, with the company also working closely with universities and training institutions to expand the semiconductor talent pipeline.

Beyond infrastructure, enterprises are beginning to see tangible productivity gains from AI, particularly in repetitive knowledge-based work. Sitaraman cited examples such as automated report generation and contract reviews, where tasks that previously took days can now be completed in minutes.

“What it does is it improves the velocity of the actions that you’re taking,” he said. “You’re not waiting for a response, you have the response, you’re waiting for what’s the next action you need to take.”

Amid growing concerns over rising token consumption and AI costs, he said enterprises are entering a new phase where efficiency will become just as important as capability.

One of the biggest shifts underway is how organisations manage context. Instead of feeding large amounts of information into models, companies are increasingly narrowing the scope of queries to improve accuracy and reduce costs.

“People are no longer going to say, dump everything into the model and see what you get,” Sitaraman noted.

The focus on efficiency comes as businesses grapple with the economics of deploying AI at scale. While the industry is still searching for answers, he expects enterprises to become far more deliberate in how they use models and consume compute resources.

As companies move from experimentation to production deployments, he said innovation will need to be balanced with governance and spending controls. “You have to have a certain budget to innovate,” Sitaraman said. “But the guardrails keep you on track.”