An AIM Research report published this month on the state of the semiconductor industry in India in 2023 highlighted some interesting findings. For instance, semiconductor companies with a smaller workforce of 101-500 had a higher market share (41%) than those with a larger headcount of 501-1000 employees (30%).
Upon further investigation, it becomes apparent that a majority of the companies with relatively smaller workforces are service-based, such as Tata Elxsi, Capgemini, Tech Mahindra, Accenture, and others. Meanwhile, companies with larger workforces, such as Siemens, Samsung, AMD, Qualcomm, and Intel, have established their R&D centres in India to develop products locally.
AI impact on chip design
The fact that a larger market share is occupied by smaller companies indicates that many companies abroad outsource chip designing to India – for reasons that include the availability of talent and cost-effectiveness. Naveed Sherwani, CEO at RapidSilicon, an application-specific FPGA design company, mentioned that they outsource many routine (junior level) jobs to countries like India, Malaysia, and China as many design centres were opened in those countries. “In most cases (not all), senior high-level architecture work is done in the USA and follow-on work on architecture, physical design and verification is done in these design centres,” he said.
However, Sherwani anticipates a significant disruption on the horizon that could reshape the global workforce landscape in chip design. The experiments and development of RapidSilicon’s AI-based tool for FPGA designers, known as RapidGPT, showed about 5 times (more in many cases) improvement in the output by senior engineers.
The experiments were mostly carried out in the front-end such as in writing RTL (Register Transfer Level) for various tasks, documenting, finding bugs, developing test benches, as well as, in running tools based on various choices and options intelligently. The tool is helping designers to optimise their solutions, increase productivity, and reduce time-to-market. According to him, these kinds of changes may be directly eliminating many outsourced jobs, maybe even entire design centres.
Workforce transformation looming
This change will result in a notable realignment of the labour force in India's semiconductor industry. According to the report, currently, 38% of the workforce is employed in electronic design automation (EDA). This is due to the abundance of skilled engineering talent in the country, as well as the cost-effectiveness of operating design centres here, making India an attractive location for this segment.
Naturally, employing AI will further reduce the cost of operations for chip companies. Furthermore, what is important to note is that the bar for passing what a machine can already do is going to rise higher with each new powerful model. And as Sherwani says, in the short run, companies with senior employees will automatically have an advantage because AI needs someone to steer them to do particular tasks.
However, in the long run, junior folks will be able to upskill faster to stay ahead of the innovation curve. According to Sriharsha Pudi, senior engineer, chip design at Marvell Technologies, companies that rely on outsourcing can only sustain by investing in employee development overall. By encouraging first principles thinking when tackling complex design challenges, employees can enhance their domain-specific knowledge of the ASICs they work on and gain a better understanding of the complex microarchitectures of chips in the long run.
In an earlier conversation with Jim Keller, CEO of the AI computing company, Tenstorrent, said they are already using AI for tasks like test generation and are experimenting in other areas as well. While he did acknowledge that it will cause a massive transformation in jobs, he also stated that it is too soon to say how many people would be left unemployed.
“It's going to destroy jobs and create jobs pretty fast. People always think the next technology revolution will destroy jobs. It's never happened, right? Is this the time? Maybe. We've never built technology that smart. We built technology that is efficient, effective, fast, has more capacity, more refinement, and more accuracy, but this time we have a technology with judgement,” he said.
Promoting India-led chip design
On a positive note, the Indian government is actively working to develop a semiconductor ecosystem within the country. The recently launched ChipIN centre, located in C-DAC Bangalore, offers an array of advanced tools for the entire chip design cycle, including front-end and back-end design, PCB design and analysis, and more, catering to digital, analogue, RF, and mixed-signal designs.
The centre aims to foster an innovation-led ecosystem for entrepreneurship and startups in India, facilitating the creation of indigenous IP cores, chips, and systems in areas such as 5G/IoT, AI/ML, and automotive and mobility. Ultimately, the goal is to establish India as a leading contributor to the semiconductor industry worldwide.
Perhaps with this, India will see a decreasing reliance on the West, and develop its own ecosystem which fosters talent, shifting from a labour-centric economy to an innovation-driven country.
Your reaction