Tech startups going for IPO should have 8-quarter rolling plan: Mohandas Pai

If the market is not very good, raising money gives you $5 billion. That's the reality of life. There's nothing wrong in that.

You want to raise capital, you can't raise it on your own. You've got to get money from somebody. They should be willing to pay the price. Everybody has to face reality. We live in the real world.

Are you expecting all the tech IPOs which are in the pipeline to go through this year?

Many of them will make it through, because many of them feel the public markets are the right place for them. There are about 32 tech IPOs which have happened. They're doing fairly well. They created a cumulative value of about $1,900 billion the last time I saw. It's very interesting.

And I think the key is very simple, they have to demonstrate value after listing. Now the key is you just find out, and I did a report which said 50 percent of them have not grown much as promised after the IPO.

So before you go for the IPO, you have to make sure that you have an eight-quarter rolling plan and a four-quarter 90 percent firm plan. And execute, execute, execute. The IPO market is there for everybody to raise money. It rewards you if you do well. It hurts you if you don't do well. That's the reality.

Face the reality and go on. It's not a negotiated market like the private markets. But once you go to the IPO market, you don't have all those share agreements that you have with investors which inhibit you, which deprive you of your rights.

It's new in the market. I think it's a good thing to do. They should grow.

Which are the startup themes you are looking to invest in 2025?

In general today, every company should have an AI layer, and then we will focus on deep tech and SaaS with AI. I think that will remove much of the barriers they have for growth. Then we should pick companies in let's say fintech and healthtech, in areas which provide technology to financial institutions, not lending.

Are there any exciting AI startups that have caught your attention?

There are many of them which are very small and coming up but we have to wait and see.  See, ChatGPT (OpenAI) is one single company, maybe two or three others which came up in the US. It's not going to happen every week. And ChatGPT came up after some 12-15 years. It did not come up yesterday.

You can't create great companies in a hurry.

How do you see the AI wave impacting the traditional IT services business models?

I keep reading articles that say next month they are going to write 85 percent of code with AI, in six months 100 percent of code. That's not the way code is written. The machine can write code, we have to test it. And you've got to have use cases. Then somebody has to observe it. They've got to manage the risk whether the code is actually right. So there has to be human activity. Perhaps if earlier 100 percent human activity was required, now it will come down to 70-60 percent.

Yes, AI will definitely assist the process. AI will definitely improve the quality, bring in certainty because the machine can be trained to repeat itself and the speed of execution will happen. All that will improve productivity.

In the past year, maybe productivity has gone up by 15 percent, and this year it may go up another 15-20 percent. You've already seen it in the numbers that the height becoming stagnant and revenues going up.

Please remember, every time a wave of technology happens, disruption takes place. On one side of the equation, there are a few disruptors; other side, there are millions of enterprises who want to use it.

Enterprises are scared. They've got accumulated software, they've got complex software, they've got all kind of intergenerational software. All that has to happen. So I think it takes time but we've got to be patient. But certainly, AI is a disruptor and AI has improved productivity by huge numbers.

Since you mentioned productivity, what will be the impact of AI on headcount and IT hiring? Job disruption seems real.

It is real but please remember, India is in a good position because there are no people in the western world in sufficient numbers to use AI. It's only in India. So they'll come to India.

Suppose a big enterprise in America wants to use AI, how will they use AI? They've got to hire new staff. They've got to hire data scientists. They've got to hire everybody. They've got to have testing. Who will do it for them? They'll come here and they'll do it.

You see, the key question you've got to ask is, is global IT spending going up because of AI? In the last two years, it's been flat. The growth has been muted.

Once people say, I'm going to invest so much in AI, then growth will happen and these people will benefit as they'll do the AI.

Many of the companies have already got vertical AI models.

What  do you think about IT companies building their LLMs?

They're already building small language models (SLMs). LLMs is difficult to build. Infosys is building, TCS, Tech Mahindra, they are all building SLMs.

LLMs are very expensive. LLMs have to be trained. So that is a very different thing.

They're not building LLMs because building LLMs to reach the market is not their job. And SLMs help them improve their revenues and improve their profits and help their customers. So they're doing that.

In which area are you going to build an LLM to compete with ChatGPT? What is the point? You spend billions to do what?

Will AI-focused SaaS players have a better opportunity than IT players amid shift to platform-based offerings of IT firms, especially for Gen AI/ AI use cases?

Yes. Because, you know, I think almost all SaaS players are doing AI. If somebody says, I am an AI-focused SaaS player, just ask others. Everybody says the same thing.

AI has become horizontal to be used in every piece of software. It is like the internet. So, you know, SaaS companies have to use that to reduce the quantum of work that happens on SaaS, to be more responsive to customers, to collect all the learnings and make the model better.