Foreword
Over the past few years, innovation in Generative AI (GenAI) has progressed at an extraordinary pace, reaffirming its transformative potential across a number of domains.
The possibilities are vast and hold the promise of profound changes on the horizon for millions of Indian citizens. Technical breakthroughs have been jaw dropping. We have quickly moved from auto-complete chatbots to reasoning machines capable of spinning out credible, human like 'Chains of Thought' (CoT) to find solutions to complex problems. Today, multi-modal large language models (LLMs) can enable seamless processing of text, audio, image, and video. Emerging trends like Agentic AI are enabling autonomous entities capable of taking actions. The evolution of new hardware platforms and new AI accelerators has ensured the computational power to support increasingly sophisticated models, having even a trillion parameters and groundbreaking efficiency.
Along the way, the cost of intelligence has fallen, driven by the open-source movement and the trend to use purpose specific small language models (SLMs). This is making AI accessible to smaller businesses and very soon it may become possible to use open-source models for as low as a few thousand rupees a month in India.
Yet, amidst all this innovation, enterprise adoption rates of GenAI remain very low. Our survey shows that 36% of Indian enterprises have allocated budgets and begun investing in GenAI, while another 24% are testing its potential. Technology sector clients are leading the way, with Life Sciences and Financial Services following suit. Despite this, the business value remains limited, with just 15% having GenAI workloads in production and only 8% able to fully measure and allocate AI costs.
This is not surprising – it takes time for innovation to be packaged and made ready for enterprise adoption. Enterprises need clarity on ROI and guarantees around issues like hallucination, data privacy and algorithmic bias as they craft their digital transformation roadmaps. Over the next few years, we expect an explosion of enterprise adoption as these issues are addressed and AI and GenAI models make their way into the enterprise mainstream.
Just as during the earlier era of mobile disruption, fintech and health-tech enterprise adoption will lead to the birth The AIdea of India: 2025 5 of AI-first companies with new business models and revamped economics. These firms will compete with digital interfaces powered by chat, voice and regional language models. Algorithms and new datasets will help drive population-scale operations. AI-driven apps will transform knowledge work.
The coming wave of change has significant implications for India.
In industries like financial services, healthcare and retail, we expect AI to reshape basic processes including customer acquisition, operations and service. Industries including IT/ITeS and BPO will undergo more dramatic changes.
Next-generation industries like biotech, advanced manufacturing and renewables will have the potential to leapfrog to AI-first business models.
Our analysis reveals that, at a macro level, the AI platform shift will impact 38 million employees, potentially driving a 2.61% boost in productivity by 2030 in the organized sector. Enterprises will need to reorient themselves rapidly to deal with this coming impending tides of change.
There will also be significant pressure on India's policy agenda. On one hand, there is the imperative to realize India's potential as the use case and data capital of the world. The focus will need to be on enhancing data accessibility and compute infrastructure, fostering AI research and innovation through initiatives like localized LLMs, and addressing challenges in responsible governance, intellectual property rights, and data protection. On the other hand, we need to address the coming potential job dislocation in the workforce by implementing aggressive skilling programs and apprentice schemes.
This report is an in-depth exploration of GenAI's current state in Indian enterprises, key trends shaping its future, and implications for Indian enterprises and policymakers. I hope you find this report valuable.
Executive Summary
The entire earth will be converted into a huge brain, as it were, capable of response in every one of its parts." This was Nikola Tesla, in 1904, predicting the impact of the radio on the world. Every generation believes it stands on the brink of transformation, fueled by the technologies of its time. Today, as we contemplate the AI era, it feels like one of those pivotal moments. On one hand, there is exponential innovation — AI's promise is vast, with the potential to revolutionize industries, redefine work, and unlock unprecedented creativity and productivity.
Breakthroughs in GenAI have been astounding, and the possibilities appear limitless. Yet, there is the critical challenge of making this transformation relevant and accessible to consumers and enterprises. For AI to truly deliver on its promise, cutting-edge innovation needs to be paired with practical applications that solve real-world problems, empower users, and bridge gaps in digital access and infrastructure.
Innovation in GenAI continues at a scorching pace
Innovation in GenAI surged in 2024, marking a transformative year for the technology.
There was rapid progress in Multimodal AI, integrating text, images, audio and video into unified models that significantly enhance real-world usability. This was particularly evident in the incorporation of these models into AI-powered phones and emerging form factors like smart glasses, enabling seamless and intuitive interactions across diverse applications.
The open-source movement gathered steam. Leading open-source large language models (LLMs) such as Meta's Llama 3 and Mistral Large set new benchmarks for performance while addressing critical concerns about data privacy and security. Simultaneously, there was a growing realization that smaller, domain-specific models could often outperform their larger counterparts in targeted tasks.
Year 2024 also saw breakthroughs in reasoning. Models such as OpenAI's GPT-4o31, and Google's AlphaProof2 achieved remarkable progress in solving complex problems across disciplines like science, mathematics and programming, consistently surpassing previous benchmarks. These advanced capabilities started to get packaged into agentic AI systems which aim to independently plan, reason, and execute tasks by dynamically leveraging tools and resources. Though still in its infancy, this agent-driven paradigm promises to fundamentally reshape our understanding of work and the way we design software systems.
Hardware innovations continued to underpin these advancements in GenAI. NVIDIA maintained its leadership with the Blackwell platform, enabling trillion-parameter models while competitors drove significant breakthroughs in AI accelerators.
To view the full article please click here.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
