'Despite tariffs, India deeply critical for Google, $15 billion Vizag investment proof of that': Google India head Preeti Lobana | Full Interview

India has been a very critical country from every strategic perspective for us. We've been in India for 20-plus years, standing with and growing with India in our digital evolution across a wide range of products and services— whether it is with Search and making information more accessible to every Indian, or Maps which are a very important part of our daily lives, or empowering businesses through Google Ads or now Google Cloud.

In all of these aspects, India is looked at as a major growth area. The vision of Viksit Bharat, the honourable Prime Minister’s vision, and the India AI Mission are all very powerful statements on where we want to go . For Sundar Pichai and Google and for us, we are here with India as we look to grow our digital economy - partnering with consumers, businesses and enterprises. We are looking to do good for India on systemic challenges in terms of how AI can help in complex areas like healthcare and agriculture. In all of these aspects, India is deeply critical for us.

Stuff will happen around. We are deeply committed. You know we made a recent announcement about investing $15 billion in our gigawatt scale data center in Visakapatnam. It's undersea cables, it's renewable energy. We announced the opening of our largest ever office in India in Ananta in Bengaluru. That’s a very large office. We recently launched the first-ever in Asia Google Safety Engineering Center in India - with a focus on privacy and safety. It is the first ever in Asia. So, we are continuing to invest. Some things happen around the ecosystem but we continue to be on our path to be with India and proceed and grow.

There are four things that are critical. The first is about how do we leverage the best AI? We deeply believe our AI is very powerful. Gemini is like the best-in-class model. How do we embed AI in products and services to make sure AI is the lever that helps India leapfrog ahead for the systemic challenges? The next step is about how do we make AI models locally relevant from the Indic languages point of view or other local contexts. For example, you know how Indian women after delivery post-natal care eat gond ke laddoo? Now any model built overseas won't have that context when you talk about post-natal maternal care. So, we build all that into our model so that it becomes locally aware in terms of Indic languages but also culturally aware. How do we do this for open-source models like Gemma is a priority so India’s vastness and complexity become an advantage and we can show to the world that this is a model that can scale to the Global South and other countries.

Part two is economic value creation. Google ads helped democratise advertising for everyone from mom-and-pop shops to the biggest enterprises to help them find more customers and revenue and we are adding the power of Google Cloud in the last few years and the Google Cloud stack for enterprises to transform for more efficiency and productivity benefits.

The third bit is about evolving our products and services for the Indian context. Think about Gemini powering search and AI overviews and AI mode. We are meeting consumers where they are: they want to have more conversations as opposed to just queries.

Lastly, and very importantly, AI is a very powerful force for good but with AI comes great responsibility. How do you develop AI in a responsible manner? How do we ensure that we are building digital trust? Privacy, safety, security - we think of these as the foundation in our products and not as an afterthought. For example, think of SynthID which helps you identify AI-generated content.

We started with foundational models like Gemini built outside India, but now we’re working hard to bring the local Indian context into them. We enrich these models with local languages — you need to bring 10-12, even 22 languages to these models — and with cultural nuances. We focus very hard on making sure that the models are truly multimodal, because in India people use voice, images, video or circle-to-search as easily as text. When we open-source them through efforts like Gemma or MedGemma, the question is how do you provide the right datasets. Google is constantly working in collaboration with institutions like the Indian Institute of Science and Project Vaani so the right datasets are available for the wider ecosystem so they can train their models on Indian data.

It is amazing to see that the government has a clear vision and it has clear pillars — whether we are talking about compute, democratising access to AI or privacy and security. In each of these pillars, we see ourselves as partners, not just with the Government of India but in enabling the wider ecosystem.

We are bringing more compute, local compute, adding TPUs (Tensor Processing Units) so industries in India can train their models, along with the infrastructure investments and skilling. AI has to be inclusive for all.

And when the Government of India talks about inclusive AI, it includes many dimensions. It's also about how you are upskilling the ecosystem so they see the power of AI and can actually use it in their daily lives.

Frankly, this $15 billion investment in Vizag is the largest ever in a data centre outside of the US. That tells you the strategic importance and criticality of India and the Vizag data centre. It is a one-gigawatt data center but it is also about the commitments we are making for clean green energy: adding power to the grid, harnessing solar, wind and other sources of power.

It’s also about deploying additional undersea cables and capacity. We are talking about faster speeds, less latency and more localisation. That is a very big statement on our part to say we believe in India. It is a strategic location and we are making that data centre expansion now. It is a fairly ambitious investment. When we talk about giga-watt scale, it is very big. It is the single largest investment of its kind outside the US. No other country has such an investment.

This is a multi-year investment and we are making sure that we are executing on that. The future will be what it will be - looking at what the demand is, how do we see the compute capacity play out and other such factors. But, Vizag is a very critical part. It is not a Tier-1 city. It is Vizag.

I think the biggest challenge, going by where we are on AI, is the responsibility part. When a technology is so powerful and scaling so rapidly, there's a huge onus on many players in the ecosystem and Google to make sure it's deployed in a manner which is safe, inclusive and responsible.

So, we are giving a lot of time and attention to how we build it into our products? How do we do red-teaming of models, where models challenge each other to come up with the best outcome? We are talking about how do we best ensure the accuracy of our models and that safety features are built into them right from the start. Not as an add-on or as an afterthought. Think about a programme like DigiKavach which aims to make people aware about fraud and possible scams.

The third part is about the checks and balances. Google built in something called SynthID: when content is created using AI, there is an invisible watermarking that happens. And then we also made a detection tool available so you can see that it has been generated using AI, even though it may be copied, changed or altered.

I am sure you heard our earnings call two-three weeks ago - that was our first quarter with $100 billion revenue. Of course, Cloud is growing very rapidly but a lot of that revenue is grounded in what we get from the Ads business on the back of Search. Those results speak for the fact that search is doing well, even with AI.

What we are seeing is people have longer queries. We have 5 trillion searches. We aren't seeing a reduction in search queries. There has been innovation in Search: when you think about people searching using Google Lens, which is very large in India or Voice, which is very large in India or Circle-to-search, where you can just circle something and search it. We are not seeing Search slow down at all. Our results reflect that Search is continuing to be robust. There's just a change in how consumers are using it and how we are meeting the moment.

In fact, what we are seeing is that now when people click on a link, they get better qualified leads. Because in AI mode, we also link the sources. Those are better qualified links. People don't bounce back. They spend more time on the website. So search is robust and healthy and we are meeting the moment by evolving search.

For me and for Google, a very essential part of a democracy is the whole publisher-media pillar. And we have always partnered with publishers, with media. With evolving consumer behaviour, we are with our publishers in providing robust tool kits Compared to anybody else with LLMs, we as Google remain very committed to sending traffic to the web. There are others who have openly said likewise. We are very committed to the web. It's just that the way people search is changing. We continue to embed links, and we continue to work with media publishers to see what are the partnerships and solutions and so the ecosystem remains healthy. The most important thing is our intent to say we are committed to the web and we want to send clicks.

That happened as the outcome of a broader conversation and discussions happening with the Government of India and other governments. We remain committed to the four pillars we laid out on working with India.

Co-creation is very critical. On the enterprise side, Cloud offers Vertex AI, where there is a foundational model that companies can build on. In the examples I mentioned earlier, customising AI for India and making it relevant locally, we see that as co-creation.

We also work very closely with the startup ecosystem. The Government of India has a strong vision for deep-tech capability, and we fully align with that. Through the Google Startup Accelerator programme, we have supported around 230–240 startups. We bring a full-stack approach: not just infrastructure, cloud credits or models, but also mentorship and our best technical expertise. That incubation is also co-creation.

Of course, we are very bullish. It is wonderful to see the vibrancy. Investment cycles may go up and down but, at Google, we take a long-term view. We should not be distracted by the noise. We are here believing in India, believing in the need for deep tech, seeing the amazing talent, making those investments, not just with dollars but with our expertise.

Think about who Google is. If you look at any of our products — Maps, Search, anything — we are about scale. Google operates at population scale in different parts of the world.

When you think about the three biggies as partners, they bring scale. We also have partnerships with different parts of the government, with the IITs, with the Indian Institute of Science, and with startups. The context is always: where is the future going?

Tatas, by the way, is a critical partner and customer. I can't say more because whenever we make announcements you will see and hear those announcements. But for us, we are always looking at where the scale is and where can we go next.

Do you see Google’s business model fundamentally shifting from advertising to more enterprise?

If you look at our earnings call, which is very public, you will see that globally — and this is true for India as well — ten years ago we were largely about ads revenue. Now you have Cloud revenue, and we are also looking at things like YouTube subscriptions and Google One subscriptions for data storage and more.

We are looking at scaling Pixel too. So. broad-basing revenue beyond ads, it is happening. At scale, which you can see in our earnings call, and the same trend is playing out in India.

India is not one India, and you can’t pick and choose. After English, Hindi has the largest scale, but when we expanded AI mode into multiple languages, we looked at Marathi, Kannada, Tamil, Telugu, Bengali and others.

For me, the biggest thing is that we are not going sequentially, saying this language first or that one. The question is: how quickly can we bring the best of AI to many languages at scale? Every Indian says, my language is important to me. So we start with one and then rapidly use the technology to expand to others. We may pilot in one, but then we scale quickly to other languages.