We released Gemini in December 2023 and there’s been a significant recognition of progress since then particularly around the Gemini family. We introduced a 2 million token context window about 6 or 7 months ago that unlocked completely different sets of use cases for technology and capability enhancements as well as significant quality enhancements. Accuracy of the models relative to others in the market has been a really big focus and investment for us.

What we’ve now started to see is an additional conversation of how enterprise value can come though agents. If you and I met six months from now or even 12 months from now, the whole conversation would be around enterprise value and all the different use cases that companies are starting to use these technologies for. I think we’re just starting that journey. So, for me, the coming 12 months is about value realization of the commercial as well as on the public sector side.

AI is lighting up every single sector to do things either more efficiently or building out completely new capabilities and experiences for consumers, even in a public and a private partnership kind of way. We’re seeing lots of interest in banking and the public sector. In India, there’s an unbelievable tech industry. Banking, telecommunications sectors are really trying to think through ‘how do I take advantage of these capabilities and introduce a genetic workflow to accelerate time to value?’ Over the next year or so, every single sector you can possibly imagine will look at this but obviously there are sectors that are much more progressive, especially tech-centric companies but some of the most interesting outcomes are coming from companies that are completely digitising and accelerating their digital transformation that may have been more traditional industries.

If I go back 12 months, there were a lot of conversations around cost takeout, straight operational efficiency and repetitive tasks. I’m seeing a lot more transition to newer conversations, that are more customer-centric. If you have higher customer satisfaction, that becomes a revenue- driving conversation, not just a cost optimisation. So, the efficiency conversation, hasn’t gone anywhere and it continues to be a very important part of AI. But what we’re seeing is a transition to more front-office, customer experience, customer journey conversations, B2B and B2B2C conversations. I think one of the biggest areas is changing customer engagement.

Within India, public sector is a key area of impact for us. On January 13 we met with Maharashtra’s Chief Minister and signed a strategic MoU for an AI Centre for Excellence where Google will partner with the State. Two weeks ago, we signed up with the Uttar Pradesh government for building an agentic framework for farmers. A lot of that work is happening out of India. Our engineering and product teams have been working on that. And then, there are other States like Andhra Pradesh. So, we have significant partnerships with various States as well as the Central government. For example, the IT Stack for India. We are also looking at various use cases around citizen, governance, re-address a lot of complaints, deviances, police and traffic management, smarter policing, golden records. We’re working with a number of states on golden records. How do you create a set of golden records around the citizens so that citizen services can then be effectively provided? All of these are using AI.

NVIDIA’s latest AI technology, Blackwell, will be available to Google Cloud customers in early 2025. This integration is set to further enhance our AI capabilities, offering more powerful tools for innovation. In 2024, we also expanded this collaboration to provide enhanced AI infrastructure and software solutions, enabling the development and deployment of large-scale generative AI models and accelerating data science workloads. In November 2024, Google Quantum AI began utilising NVIDIA’s CUDA-Q platform and the NVIDIA Eos supercomputer to simulate qubit designs. This collaboration aims to accelerate the development of practical quantum computing devices by enabling large-scale simulations that inform next-generation quantum processor designs.

That was more in context of the cyber space, where AI can be used by the good guys but also by the bad guys. We have a lot of work that we continue to do with the regulatory bodies and different nation states to make sure that our capabilities are being used for a force for good in line with some of the regulatory requirements of those countries.

We believe it is important for these kinds of technologies to have regulation but that needs to be supported by a pro-innovation agenda. We need to also develop global frameworks, but also make sure that we are partnering with different nations that may deviate from some of those frameworks to make sure that we can do business to support their citizens as well as the enterprises. So, it’s an ongoing set of challenges.

Through the next 6-18 months, we’re going to see a significant uptick in enterprise adoption of these kinds of capabilities. But they only really become unleashed when people start to build out these agentic capabilities, because that’s the mechanism for scale as well as those kinds of capabilities and natural language interactions, that’s where you’ll see business users start to get materially more value versus more traditional consumer-based applications, which is where a lot of AI has permeated in the early adoption phase.

We view the importance of being able to be best in class capability as well as best in class for commercial structures at the same time. We think we have a unique advantage of being the best model at the best price point, which is really important as you head towards the enterprise since they will have different requirements and cost sensitivities. That’s a really important underpinning that you’ll continue to see part of going forward.