The battle for global AI leadership is often pitched as a two-horse race between the US and China, thanks to their formidable research capabilities. But new analysis of code-sharing platform GitHub shows that when it comes to writing AI code, India has eclipsed even the US. This reflects an impetus to upskill India’s vast technology talent base that was initiated by the country’s IT services companies. But the recent evolution of India’s economy means they will not be the only ones to benefit.
A country’s AI power is usually measured by the volume and quality of its research. The US has dominated AI research since its inception, although this lead in research quality has recently been challenged by both China and the EU. But new analysis by the OECD of data from Microsoft-owned code-sharing platform GitHub reveals another contender in the AI race: India.
The data is gathered from public AI-related code repositories, or repos, that are hosted on the platform. Examining the location from where contributions, or 'commits', to these AI repos are made reveals that in 2019, India overtook the US as the principal source of AI-related code. In 2020, it accounted for 30% of all commits, double the US figure. (Chinese developers may be underrepresented on GitHub - last year, the country launched its own equivalent - but they are not absent, the OECD says).
The data suggests that India's AI prowess has been underestimated, says Luis Aranda, AI policy analyst at the OECD. "We always talk about the two main players. But India is emerging as a third, so perhaps it's time to turn our attention in that direction as well."
How India built its AI talent base
The GitHub data comes as no surprise to Saurabh Gupta, president of research and advisory services at analyst company HFS Research. It is the result of a concerted effort to ready India's vast talent skills-base for the AI era, he says, initiated by the country's IT services giants.
"When AI, analytics and automation started to appear around 2010, India['s IT service providers] were somewhat threatened, because they felt they would cannibalise the whole offshore delivery model," he says. "But within three or four years, they started to embrace that change. You started to see all the Indian majors doing huge amounts of training and you started to see this AI [talent] population expand."
Industry bodies, including the influential IT services lobby Nasscom, followed suit and AI became an educational priority for the country. "Universities, engineering schools and even high schools now incorporate AI in their training," says Gupta.
India lags behind the US and China in cutting-edge AI research that moves the field forward, says Husan Chahal, research analyst at Georgetown University's Center for Security and Emerging Technology (CSET). This is due to a relative lack of PhD students and the supercomputers needed for high-end AI research.
In terms of getting economic value [from AI], and who is doing the crux of the work, India has a huge role to play.
Husan Chahal, CSET
But when it comes to developing basic AI research into technology solutions, Chahal says, India more than holds its own. Analysis by CSET shows that four US corporate AI powerhouses (Facebook, Google, Microsoft and IBM) have five AI labs based in India between them. All five are involved in experimental development of AI solutions. "There are issues with the PhD talent base but in terms of getting economic value [from AI], and who is doing the crux of the work, India has a huge role to play," Chahal says.
Indeed, the rise of India's AI capabilities reflects a maturing of the technology beyond its highly specialised research phase, in which China and the West dominate, into commercial application, says Gupta. "AI isn't horizon three," he says, referring to the furthest boundary of innovation. "It's horizon one."
Who will benefit from India's AI talent?
Ten years ago, the main beneficiaries of this AI talent boom would almost certainly have been India's IT services giants, such as Infosys, TCS, and Wipro. Today, the story is not so simple.
This is partly because Western services firms have all increased their Indian footprints, Gupta says. "Whether its Accenture, IBM, KPMG, or EY, their Indian workforces are now as big as the so-called Indian companies," he explains. This means a skilled AI developer in Bangalore is just as likely to join a US-owned company as a domestic one. "India is no longer a silo," Gupta says. "It's a mesh of the global talent pool."
Perhaps more significantly, though, the IT services industry is no longer the only career path available to India's technology talent. "India's younger population is realising, 'we can do more than just [outsourcing]'," says Gupta. This has been catalysed by the pandemic, he adds. "The attrition rate for some of the leading IT services providers is going through the roof, from 10-15% [annual turnover of staff] to 30% or in some cases 50%."
India's younger population is realising, 'we can do more than just [outsourcing]'.
Saurabh Gupta, HFS Research
Some of this technology talent has joined the gig economy, Gupta says, through platforms such as TopCoder. "When we talk about crowdsourcing, we often think of low-end, commodity work," he explains. "But [TopCoder] is on the other end of the spectrum. They are crowdsourcing really complicated stuff, like data science and AI research, for clients including NASA."
Meanwhile, India's start-up sector has enjoyed an explosion in investment. In the past 12 months, India's start-ups attracted $28bn in venture capital, according to data from Crunchbase, fourth behind the US, China and the UK. AI start-ups have been especially well-funded: in 2019, AI start-up investment grew faster in India than in any other market, according to CSET. Notable examples include virtual assistant provider Haptik, which was majority-acquired by Indian industrial giant Reliance in 2019.
These start-ups offer appealing careers for a generation that wants to "build something," says Gupta. More so than previous technology paradigms, therefore, India's AI talent boom could fuel the country's domestic digital sector.
Nevertheless, Gupta says, services will always play a part in India's tech-driven economy. "What India does well is provide talent at scale," he explains. As AI becomes mainstream in the coming decade, India's depth of talent will be its superpower. "That's going to be India's competitive differentiation."