With 300,000 employees using Microsoft 365 Copilot in India, what does this mean for enterprise AI adoption?
Copilot is now becoming the connective layer across more than 300,000 engineers and associates at Infosys, TCS and Wipro.
For me, this means three things.First, Copilot is getting embedded in how Indian IT services delivers to the world. It is becoming the connective tissue or the user interface for AI in Indian IT services, which is very exciting.
Second, this is proof of enterprise AI at scale. We are way past experimentation and pilots and have moved into large-scale, sustained deployments.
If you look at usage statistics, Infosys has 91 per cent usage, TCS has 86 per cent daily usage, and Wipro has built 29,000 citizen agents, where employees are building agents that are transforming processes.
Third, India is leading the pack. India is becoming the global template for doing AI at scale. We have the path from pilots to production.
How does this compare with global markets? Do you think adoption will now become smoother?
It has only been two years. It is amazing how impatient we become with technology.
Globally, as seen in our last quarterly results, we have reached 20 million Copilots, with 50 per cent year-on-year (Y-o-Y) growth. The scale is massive. We are also seeing that customers who use Copilot at scale continue to expand usage. The number of customers with more than 50,000 Copilots has quadrupled. Once organisations start using it and see impact and value, they scale rapidly.
The same happened with Infosys, TCS and Wipro. They were all at around 50,000 Copilots six months ago. They have doubled that number because they saw value and impact.
In terms of lessons learnt, the first requirement is leadership vision and bold ambition. Leaders must lead from the front, ensure employees have the right tools, and focus on real value rather than hobby projects.
Second, organisations need to re-architect processes. Applying AI to a poorly-designed process simply amplifies inefficiency.
Third, organisations need forward-compatible infrastructure. Poor data leads to poor AI outcomes. There is a lot of discussion around managing tokens, token factories and token ledgers, but ultimately it is about getting infrastructure right.
Fourth, this is a cultural renaissance. Organisations need to think about work charts. AI and humans will work together in different ways, and companies need to rethink how they scale.
Is the conversation around Copilot now moving beyond productivity? What are enterprises looking at today?
Initially, the AI thesis was centred around productivity and efficiency—doing things faster and more smoothly. However, in our recent Work Trend Index report, one of the most encouraging findings was that 49 per cent of Copilot work today is cognitive work. People are using it for analysis, critical thinking and problem solving. They are not outsourcing their thinking; they are using AI as a cognitive amplifier.
Another important finding is that 58 per cent of people say they can now do things they could not do 12 months ago. AI is helping build entirely new capabilities.
AI is lifting human ambition and capability. The real question is whether organisations have the leadership, culture, process redesign and structures necessary to extract value from this increased capability.
Following the Nayara incident and broader geopolitical uncertainty, conversations around digital sovereignty have intensified. Are customers asking more questions about data residency and resilience?
It is absolutely fair for customers to ask for data residency and local data capture. That is one reason we are investing $17.5 billion, on top of the $3 billion already announced, bringing our total planned investment in India to $20 billion over the next four years. The Hyderabad region goes live next quarter.
We already have the largest Cloud capacity in the country and continue investing to meet India’s needs. We offer sovereign solutions, sovereign public Cloud and sovereign private Cloud offerings. We have also made digital commitments to ensure we meet India’s sovereignty requirements.
Our focus in India is scale, skilling and sovereignty.
Can you give an update on the Hyderabad data centre? Also, how much of the demand are you planning for AI and generative AI workloads compared to traditional Cloud adoption?
The Hyderabad region goes live next quarter. It is one of the largest data centre investments in the country and will further strengthen our AI and Cloud capacity.
It is both Cloud migration and AI workloads. There is a significant wave of migration to the Cloud because organisations realise it is difficult to run AI effectively on-premises.
At the same time, AI workloads are growing rapidly, whether through Copilot adoption or broader agentic transformation initiatives. So, demand is coming from both Cloud adoption and AI.
You have said that AI will change jobs rather than eliminate them. Given the layoffs happening across the industry, what are you hearing from customers?
I want to clarify that I did not say AI will not affect jobs. I said AI will change the colour of every job.
There are three implications. First, many new roles are emerging. At Microsoft, we now have roles such as agent orchestrators, AI workflow designers and forward deployed engineers. These roles did not exist 12 months ago.
Second, every job is changing. Job titles matter less than the skills required to perform the work.
Third, AI will automate a lot of transactional work. If someone’s role is entirely focused on repetitive tasks, they will need to continuously upskill and adapt. AI lifts human capability by taking away low-value transactional work. The challenge is ensuring people have the skills required to move into higher-value activities.
Mythos has emerged as a major talking point among enterprises globally. Are you seeing customers reassess their AI governance and cybersecurity frameworks in response to these developments?
There are two ways to think about this. The first is security for AI—ensuring AI systems themselves are secure. The second is AI for security—using AI to strengthen cybersecurity. The entire physics of cybersecurity is changing.
What we have learned is that using multiple models together often produces significantly better security outcomes than relying on a single model. You will hear more from Microsoft at ‘Build’ conference.
There is concern, but there is also strong interest and commitment from customers. Organisations want to ensure they are on the right side of the equation and use AI to strengthen security.
I am saying multiple model strategy as no company today will bet everything on one model. New models emerge constantly. That is why Azure Foundry supports 11,000 models. The goal is to provide model diversity, the right model for the right use case, at the right price point and in a secure enterprise environment.
Ultimately, models will not be the moat. Organisational intelligence and the ability to use these models effectively will matter more.
Token expense is hitting companies hard. Many are shifting their usage of frontier models. How should companies weigh their expense vis-à-vis AI roadmap?
On tokens and cost escalations, I go back to the four things I mentioned. One of the things was this whole piece around forward compatible infrastructure, which is you need a token Ledger in your organization.
You need visibility to tokens being spent by employee. There are couple of things that we are doing on this journey. First is we are building Agent 365, which is a layer which is a control plane for you to manage all the agents in your organization. Second, is we are bringing multiple features, there are several flexible pricing models. So there’s a fixed user per user per month pricing model, there’s a token based consumption pricing model for AI workloads.
Again, making sure that customers have the right, have the flexibility in terms of what model they would like to use. In some cases, CFOs will develop their forecast, in some cases they will go with consumption, but make sure we bring the right models to them.
Source link
#India #global #template #scale #Puneet #Chandok
