Last Updated:
Despite the surge in AI pilots, many enterprises remain stuck at the experimentation stage. Proof-of-concept projects exist but scaling them into full-fledged systems is complex
Many companies measure AI success through model accuracy, tool usage. Experts point to a disconnect, as executives expect revenue growth while very few firms can link AI initiatives directly to profit.
The Maharashtra government’s approval of a Rs 10,000 crore artificial intelligence (AI) policy signals a shift in India’s AI journey from vision to execution. The policy, which aims to create 1.5 lakh jobs by 2031 and embed AI across sectors such as governance, agriculture, and education, also focuses on building infrastructure, supporting start-ups, and accelerating adoption among small and medium enterprises.
Key elements of the plan include access to 2,000 GPUs for start-ups, a Rs 500 crore venture capital fund, and a 20% subsidy for around 5,000 MSMEs. The policy also stresses ethical AI frameworks and local language innovation, particularly in Marathi, pointing to a more inclusive approach to technology adoption.
It is one of the most comprehensive state-led AI initiatives in India. But it also raises a key question: is India ready to move from AI ambition to scaling projects? What do enterprises lack when it comes to scaling AI?
How Rapidly Is India’s AI Ambition Growing
India’s AI ecosystem has expanded significantly in recent years, with over 5,000 AI start-ups, and a market that is expected to reach $17 billion by 2027 — growing at 25% annually. At the policy level, both the Centre and states have begun investing in compute infrastructure, skilling programmes, and innovation ecosystems.
India’s AI mission includes building an ecosystem with an estimated cost of Rs 10,300 crore-plus, aiming to expand computing capacity to 200,000+ GPUs (Graphic Processing Units), establish indigenous AI models and train 10 lakh youth.
The country ranks among the top 25% of countries globally in AI readiness, according to the Boston Consulting Group. With the right policy frameworks and resource allocation, India and India’s states have the scope and popularity to further strengthen AI adoption.
At present, 21 states currently have an AI policy, among them Andhra Pradesh, Telangana and West Bengal offer specific elements, such as AI start-up and capital support, research and talent, and outcome-based contracts, for effective implementation of state-level AI policies in the country.
States like Maharashtra are now competing to position themselves as AI hubs, offering incentives, infrastructure, and regulatory clarity. This aligns with India’s broader digital transformation push, which includes initiatives in fintech, health tech, and public digital infrastructure.
However, while policy momentum is accelerating, the real test lies in execution—particularly within enterprises.
What Are The Challenges In AI adoption By Enterprises?
At present, around 47% of enterprises in India use multiple AI use cases live in production, according to an EY-CII report ‘The Aldea of India: Outlook 2026’. PIB says around 89% of new start-ups launched recently in the country use AI in their products and services.
Despite the surge in AI pilots across industries, a large number of Indian enterprises remain stuck at the experimentation stage. Proof-of-concept projects are common, but scaling them into full-fledged, production-ready systems is far more complex.
An IBM report published last year highlights five key challenges in AI adoption, which include concerns about data accuracy or bias, organisation lack access to sufficient propriety data, inadequate AI generative expertise, lack of financial justification to start Gen AI initiatives, and privacy and data governance.
Many organisations struggle to integrate AI models into their existing systems. Legacy IT infrastructure, fragmented data sources, and lack of real-time processing capabilities often prevent AI from delivering measurable business outcomes.
For example, deploying AI in customer service, payments, or insurance claims requires seamless integration with backend systems, data pipelines, and decision engines. Without this, AI remains an isolated tool rather than a core part of operations.
Is Deployment The Real Bottleneck?
India produces a large pool of engineers and data scientists, and global tech companies continue to invest heavily in Indian AI capabilities, but the real challenge is deployment.
According to Infosys co-founder Nandan Nilekani, the rapid pace of AI innovation means technology is advancing faster than enterprise deployment, creating a widening gap between model capability and real-world implementation.
Many organisations still measure AI success through model accuracy, tool usage, or localised efficiency gains. Experts point to a disconnect, as executives expect revenue growth or margin impact, while very few firms can link AI initiatives directly to profit and loss outcomes at the enterprise level.
Building an AI model is only the first step. Turning that model into a reliable, scalable system that can handle real-world complexity is significantly harder. This requires robust infrastructure, continuous monitoring, and the ability to adapt to changing data patterns.
Enterprises also need to ensure reliability and accuracy. In sectors like finance or healthcare, even small errors can have serious consequences. This makes organisations cautious about deploying AI at scale.
MSMEs And The AI Scale Challenge
Maharashtra’s policy places a strong emphasis on supporting 5,000 MSMEs, offering 20% subsidies to encourage adoption. This is crucial, as MSMEs account for nearly 30% of India’s GDP and employ over 110 million people.
However, these businesses often lack the technical expertise and infrastructure needed to deploy AI effectively. While incentives can lower entry barriers, scaling AI within MSMEs requires accessible tools, simplified platforms, and hands-on support.
For most MSMEs, the practical route to adopting AI lies in integrating it into existing workflows rather than treating it as a large-scale transformation requiring new teams, high costs or complex system overhauls. This is why adoption is expected to be driven by plug-and-play solutions; tools for AI-powered customer support, assisted marketing and content creation, demand forecasting, inventory management, basic analytics, and computer-vision-based quality checks that improve efficiency while reducing waste, a report by the Observer Research Foundation (ORF) said.
Evidence from a NASSCOM-Meta collaboration shows that many tech-enabled MSMEs are optimistic about AI’s potential to drive growth and improve productivity. At the same time, it highlights key barriers to adoption, including limited awareness of suitable tools, affordability concerns, and budget constraints that hinder sustained use.
This shifts the policy focus from simply promoting AI to actively shaping the market by reducing the risks for early adopters, improving access to reliable solutions, and ensuring that initial experimentation does not turn into a financial burden.
Why Compute Power Is Key To AI Infrastructure?
In India’s digital economy, GPUs are increasingly being seen as the “new oil”. While a computer’s CPU functions like a single expert handling one complex task at a time, a GPU works more like a large team solving thousands of smaller calculations simultaneously. This ability, known as parallel processing, is crucial for AI systems, enabling them to analyse vast datasets quickly and learn patterns that power real-world applications.
The allocation of 2,000 GPUs under the Maharashtra’s AI policy highlights compute power. AI models, especially large language models, require significant computational resources. Access to GPUs and cloud infrastructure is essential for training and deploying these systems.
While historically lagging, massive commitments, including a 200,000-GPU target, signal a rapid shift towards sovereign AI infrastructure.
In February, India announced an expansion beyond its existing 38,000 GPUs, aiming to deploy over 50,000 additional GPUs within six months, with a long-term goal of 200,000+.
Without widespread access to affordable compute, AI adoption will remain uneven.
What Comes Next For India’s AI push
The success of Maharashtra’s AI policy will depend not just on funding, but on execution. Creating jobs, building infrastructure, and supporting start-ups are important steps, but they must translate into real-world applications.
For India as a whole, the next phase of AI growth will be defined by operationalisation, moving beyond pilots to production systems that deliver tangible value.
This will require collaboration between government, industry, and academia. It will also demand a shift in mindset, from viewing AI as a tool to treating it as a core business capability.
India’s AI journey is at a turning point. The country has the talent, the market, and now increasingly, the policy support to become a global AI leader.
But the missing layer — deployment, integration, and real-world execution — remains a critical gap.
Stay Ahead, Read Faster
Scan the QR code to download the News18 app and enjoy a seamless news experience anytime, anywhere.
login
