IN the 1960s, Avis, a car rental company, ran a memorable advertisement that read: “We are only No 2. We try harder." It converted an also-ran status to its advantage. This should be India’s motto as we seek to join the artificial intelligence (AI) superrace after a late start, a race where the US and China have a headstart over the rest of the world.
First, the good news. In emerging technology areas, there are advantages to not being first or No 1. Tech giants tend to rush into frontier areas for fear of being left out of the moolah. But this comes with the danger of sinking billions, even trillions, of dollars that may ultimately be reduced to ashes if the technology does not deliver as promised, and/or if strong regulatory headwinds kick in. The purpose of any innovative technology is to deliver returns on the capital invested and not just produce a shiny new toy that looks cool.
Second, in frontier technology areas like AI, where the immediate threats of job losses and spikes in cybercrime are significant, the social and political pushback will be stronger than ever once it begins. After AI tools made coding skills less important, India’s big tech services companies like Tata Consultancy Services (TCS) have either cut jobs or reduced the pace of hiring new talent. In the US, jobs have been cut by almost all major tech giants, many with trillion-dollar market valuations: Google, Meta, Microsoft, Amazon, Oracle, etc.
Not all of it is due to AI, but the intensity of competition and the rush to invest in AI have surely led to the need to find savings elsewhere. Even in China, where it is presumed that worker unrest will not happen as the Communist Party holds all the levers of power, a New York Times report (“China’s plans to save jobs from AI") says that AI deployment on a scale where it impacts jobs is a hugely sensitive issue. So, rest assured, AI is not going to remain the unregulated Wild West it currently seems to be anywhere.
Third, some instances of FOMO (fear of missing out) about India are misplaced. In May and June, when stock market frenzy pushed up AI and AI-related software and hardware companies’ share prices to stratospheric levels (chips, semiconductors, large language model companies, etc), India’s rank in terms of market capitalisation fell behind South Korea and Taiwan. As the frenzy abated last month, India is back at No 5.
It is worth noting that a diversified market is safer. Countries that put all their eggs in one basket have the most to gain when the going is good and the most to lose when it isn’t. Taiwan, where semiconductor company TSMC accounts for more than half the country’s market value, rose and fell with the semiconductor frenzy. The Netherlands, Denmark, South Korea, Singapore, Thailand, Indonesia and Spain are countries whose leading stock accounted for more than a fifth of the total market value, as represented in the Morgan Stanley country indices.
In India, where we constantly make biting comments about Adani and Ambani, the top three stocks in the Morgan Stanley index represent less than 18% of the overall market. We are diversified, and thus safer from excess volatility in one or two stocks.
Fourth, AI is yet to show that it can deliver returns on investment in many cases. To understand why, we need to explain the word “token", which is the basic unit on which companies producing AI tools and software charge their clients. A token is a small unit — often a part of a word — and AI processes requests by breaking down your words and sentences into tokens. Since every token used takes up a lot of computing power in large language models like ChatGPT or Claude Code or Google Gemini, when users are billed on the basis of tokens, they end up paying a lot of money for projects that may be from delivering higher efficiencies or saving costs.
Uber found, to its horror, that its Claude Code users gobbled up the entire year’s AI budget in four months. Even Google, Microsoft and Accenture have placed limits on how many tokens their employees can consume. Uber has put a limit of $1,500 per coder for employees using Anthropic’s Claude Code. Clearly, AI is moving beyond just availability to concerns over how much it costs before it delivers benefits.
Fifth, the previous point on tokens provides us a segue into why it is too early to write off India’s software services companies. All of them have lost massively in terms of market valuations as AI tools are seen to be reducing the need for more coders to do client work. With AI tools now widely available, clients will be demanding that projects be priced more in terms of costs and benefits, and not the number of coders employed.
While this is certainly a worry, and software services companies have to quickly transition from labour arbitrage-based contract pricing to overall pricing, which involves both AI tools and people, India still has an advantage. US and global companies, which have many legacy systems, and where a TCS or Infosys or HCL Tech would have done work in the past, will still need these very same companies to deploy AI by stitching together a new system using AI tools and labour. At the recent Infosys annual general meeting, Chairman Nandan Nilekani refuted the idea that companies like his will be swamped by AI. He said that Infosys was already collaborating with 90% of its top 200 clients on their AI transformation journeys. He believes that the adoption of AI will open up a new opportunity of over $300-400 billion in AI-related services by 2030.
The nuanced conclusion is that India will face challenges as it has not rushed into AI. It has missed the bus for now, but given that AI itself is going to face headwinds as return on investment calculations and job losses loom larger, there is a window of opportunity for those who adapt faster now to AI. The first, expensive bus which we missed, will not cost us too much, but we cannot afford to miss the next bus, which is about adapting to the new AI environment both for use inside our own businesses, and for servicing clients.
The key element in the Avis advertisement is the second line: “We try harder". It is okay not to be No 1, but we must try harder.
