'AI isn't just..': Anand Mahindra backs Palantir CEO's theory, sees India's edge lie elsewhere

As countries and companies pour billions into building ever more powerful artificial intelligence models, Anand Mahindra believes the industry's biggest long-term advantage may lie elsewhere. 

Reacting to a recent interview by Palantir CEO Alex Karp, the Mahindra Group chairman argued that while frontier AI models remain important, the real commercial and strategic edge will increasingly come from the application layer — the software built on top of those models that makes AI usable, secure and valuable for businesses. 

What Anand Mahindra said 

In a post on X (formally twitter), Mahindra said Karp's interview would "be widely discussed and debated," acknowledging that some may view the Palantir chief's comments as self-serving. However, he said the underlying framework deserves serious consideration. 

According to Mahindra, AI consists of three interconnected layers: 

• Compute — the computing infrastructure powering AI. 
• Models — the large language models that generate intelligence. 
• Applications — software that turns those models into practical enterprise solutions. 

Mahindra agreed with Karp's argument that critical infrastructure cannot rely on AI models alone. 

"Critical infrastructure doesn't run on a model alone; it needs an application layer sitting on top," he wrote. 

'AI is like a smartphone'

Drawing from his recent shareholder letter to Tech Mahindra, Mahindra compared today's AI landscape to the smartphone revolution. "A smartphone is remarkable technology," he noted, "but it became indispensable because of the apps and experiences built on top of it." 

Similarly, he argued, lasting value in AI will not necessarily be created by the companies building the underlying models or chips, but by those developing specialized applications that solve real-world business problems. 

"The ecosystem determines who creates lasting value, not the chip or the model underneath," he said. 

Why the application layer matters 

Mahindra highlighted Karp's argument that enterprise AI requires safeguards beyond what a foundation model can provide. 

According to Karp, the application layer makes AI "safe and precise" by addressing questions that businesses and governments cannot ignore, including: 

• Who owns the data? 
• Where is the data stored? 
• Are prompts secure? 
• Is information being transferred externally? 
• Can decisions be audited and governed? 

These capabilities become especially critical in sectors such as defence, classified government programmes, banking, healthcare and other highly regulated industries. 

India should still build frontier AI models 

Despite emphasising applications, Mahindra clarified that he continues to support India's efforts to develop sovereign frontier AI models. 

However, he suggested that if AI models increasingly become commoditised — with multiple open and proprietary models available — the country's strategy should not focus on models alone. 

Instead, he argued that compute, models and applications should evolve together, allowing India to build a complete AI ecosystem rather than relying on any single layer. 

Why AI service companies could benefit 

Mahindra believes AI service firms may enjoy a durable competitive advantage because they already possess decades of enterprise workflow expertise. 

Rather than owning the underlying model, these companies can remain model-agnostic — choosing whichever open or proprietary model best fits a customer's needs — while building secure, industry-specific applications that preserve business knowledge, governance and operational control. 

"The application also has to enable business enterprises to preserve their 'alpha'," Mahindra wrote, adding that this is where AI service companies are best positioned to create long-term value.