Swedish industrial AI deal exemplifies Indian software fix to AI threat

TCS and ABB’s changing relationship

The relationship between TCS and ABB has evolved similarly in recent months. In December, they reset the contract under which TCS has supplied IT cloud services to ABB since 2020. TCS would use AI to automate its own services under the contract, but it is now being employed to consolidate ABB’s IT infrastructure as well to make “a stable foundation for innovation”. They went on to announce a strategic partnership in March to collaborate on developing industrial AI systems, the pursuit of which requires good data.

TCS billed the partnership as a pairing of ABB domain expertise in electricity and industrial automation with TCS’ expertise in computing. That meant that TCS would get access to ABB domain knowledge and data to build industrial AI systems, as well as its customers to sell them, said Anupam Singhal, president of manufacturing at TCS.

But the pair also intended to co-develop industrial AI and sell it to TCS’ industrial customers, he told Computer Weekly. TCS would help to marshal the data their industrial AI systems would need to operate, which resided in the operational technology – the plant machinery and control systems big industry uses. It would govern cyber security of data gleaned from industrial control systems that have historically worked in isolation, and is now vulnerable to hackers as it is exposed to the world by being connected with IT systems. The pair are working to change culture in big industry that was opposed to AI, he said.

For India’s IT services industry to survive AI automation, it should focus on building AI systems to serve specific economic sectors, according to the National Institute for Transforming India, the government policy think tank. Meanwhile, India’s Ministry of Finance said in January that India lacks the energy, capital, compute and data infrastructure to build frontier generative AI (GenAI) models capable of competing with the US.

But India also has many strengths, with more published AI researchers in academia and industry than any country in the world except the US, and a workforce whose basic AI skills rank India ahead of any other country. It also has vast datasets in sectors where it is strong, such as health, farming, education, government and finance. The government wants those to become the basis for sector-specific AI systems, according to government plans.

Some of the infrastructure to do that will be provided by a 100 MW datacenter that TCS agreed to build with OpenAI in February, with plans to scale it tenfold. It would make India a global AI hub, said TCS parent, the Tata Group conglomerate.

TCS established a company for the purpose in October. It would aid a TCS plan to become the world’s largest AI services firm. The strategic partnership gave ABB an opportunity to supply the datacenter energy systems, said Singhal.

There has been a flurry of similar AI partnerships between consulting and industrial firms in recent months, as each seeks to get the domain knowledge or computing skills it needs to develop industrial AI. TCS itself has struck strategic partnerships with ABB competitors Honeywell and Siemens Energy. The Siemens deal is similar to the one TCS struck with ABB, only it was more explicit about the role the German firm will play in building HyperVault datacenters. Siemens has also partnered with TCS rivals Accenture and Capgemini.

ABB, which spent $1.3bn on research and development (R&D) last year, already has a global research centre in Bengaluru, an Indian techno-industrial ecosystem likened to Silicon Valley. It has some 2,500 staff there, and about half of 7,800 staff it employs worldwide on R&D work on “digital, AI and software”, but only 300 on AI specifically. It has 250 AI projects in progress.

ABB declined to comment on what it gets from the TCS partnership that it does not already do itself, nor would it state what hopes it has for participating in datacentre construction in India. But datacentres have been a booming business for ABB and other energy infrastructure firms.

Other notable deals Modi did in Northern Europe was to secure Norwegian help in building digital government services, Finnish help building expertise in quantum, and Dutch help building an advanced semiconductor chip fabrication plant.

India’s government AI reform strategy

Indian IT services firms have meanwhile adopted the government AI reform strategy, as evidenced in public statements, echoing its concerns about the AI threat and how it should be made opportunity.

Tech Mahindra, India’s fourth largest IT services firm, is using experience it gained building native language Indian AI models to build systems focused on specific industries, said Sham Arora, its chief technology officer (CTO). But it is doing that only for those where it has most experience doing IT services, as building a specialised AI system requires access to domain knowledge and data.

“This work has to have domain expertise and it has to have client data, so you need a client relationship,” said Arora.

The domain knowledge and trusted customer relationships that Indian IT services firms have built in sectors that have outsourced their back-office IT operations to the sub-continent in the past 40 years has become valuable to India in other ways, said Radha Basu, CEO of iMerit, which helps US frontier AI firms to train their models to operate in specialist areas such as healthcare.

Basu stated that big US AI firms that want to sell their systems into industry need the knowledge to build specialised AI systems and trusted relationships to sell them. That was also the justification OpenAI rival Anthropic gave for a partnership it struck with TCS rival Infosys in February – Anthropic would collaborate with Infosys, which has domain expertise and AI models, to develop applications for specific industry sectors.

“The frontier labs are still largely product companies and they don’t have long-term relationships with anyone. How do you move these models into the enterprise, where workflows, trust, culture and long-term customer relationships matter?” said Basu.

“The Indian IT and services companies have a real opening because they’ve been working with insurance companies and healthcare and banking and even manufacturing companies. That trusted enterprise relationship is critical for the large model companies.”

Basu added that India has vast niche expertise beyond IT services that the US AI firms need as well, along with PhDs in disciplines such as mathematics and sectors such as health. If the frontier firms combine that with the domain expertise of the IT services firms, it “will really scale the adoption of AI in a large way”, she added.