The Architect of Enterprise AI
In 2015, Vishal Sikka convinced Infosys to donate $3 million to an obscure nonprofit called OpenAI. Sam Altman was running a research lab, not a product company. The public had no idea what a large language model was. Most corporate boards would have laughed the proposal out of the room. The Infosys board eventually did - when Sikka later tried to get them to invest $1 billion. He left the company two years later. The nonprofit became the most talked-about company on Earth.
This is the recurring pattern with Vishal Sikka: he sees something real, moves toward it, and the rest of the world catches up later. He was doing it with in-memory computing in 2008, with enterprise AI in 2019, and he is still doing it today - this time by insisting, loudly and in public, that generative AI as currently sold is mostly theater.
Born in Shajapur - a city of roughly 100,000 people in Madhya Pradesh - Sikka grew up in a family that valued education as infrastructure. His father was a railway officer, his mother a teacher. By the time he was a teenager in Vadodara, Gujarat, he was reading essays by Marvin Minsky and Joe Weizenbaum, the AI pioneers who were debating whether machines could think at all. He wrote Minsky a letter from India. Minsky wrote back. When Sikka eventually applied to Stanford, Minsky wrote his recommendation letter.
That kind of directness - reaching out across the world to the person who actually matters - would define his career. He completed his B.S. in Computer Science at Syracuse University, then his Ph.D. at Stanford in 1996, with a dissertation titled "Integrating Specialized Procedures into Proof Systems." Before corporate life, he founded iBrain (acquired by PatternRX) and co-founded Bodha.com, which focused on enterprise application integration without middleware dependencies.
"Generative AI is inherently hallucinatory - not a deterministic system, but a system based on probabilities. You pick the most likely next token when you give it a prompt."- Vishal Sikka
SAP came calling in 2002. By 2007, Sikka was SAP's first-ever Chief Technology Officer - the company had existed for 35 years without one. He used the role to do something audacious: he challenged the fundamental architecture of enterprise software. Databases had always lived on spinning disk. What if they lived in memory? The result was SAP HANA, an in-memory data platform that processed in minutes what previously took days. It became the fastest-growing product in SAP's history. It also changed what enterprise software was capable of doing.
In 2014, Infosys - India's second-largest IT services exporter and a company historically led only by its founders - handed Sikka the CEO title. He was the first outsider ever to run it. The mandate was transformation. Infosys was a $10 billion business built on labor arbitrage at a moment when automation was threatening the entire model. Sikka launched what he called "Renew + New" - cut costs through automation while simultaneously investing in AI, design thinking, and new service models. He brought in a design team. He talked about amplifying human potential rather than replacing it. He invested in OpenAI while it was still a nonprofit research collective. Three years in, he was out. The circumstances of his departure were tangled in founder politics and disagreements over strategy. He has described the period diplomatically.
In 2019, he started again - this time entirely on his own terms. Vianai Systems launched with $50 million and a thesis that the enterprise AI market was about to explode, but that it needed trustworthy infrastructure before any serious company would bet their operations on it. By 2022, Vianai had raised a further $200 million, bringing total funding to $440 million. The company is based in Palo Alto and its advisors include Indra Nooyi, Sebastian Thrun, Alan Kay, and John Etchemendy.
Sikka's philosophy on AI has grown sharper and more contrarian as the hype has intensified. At the India Today AI Summit in February 2026, he dismissed talk of artificial general intelligence as "nonsense" - AI, he argued, is a tool, and should be evaluated like one. He has described the current generation of generative AI products as "inherently hallucinatory" - probabilistic systems that pick the most likely next token, with no inherent relationship to truth or reality. His company's approach is to build AI that can be interrogated, explained, and trusted - especially in environments like financial services, manufacturing, defense, and aerospace, where a hallucinating AI is not just unhelpful but potentially dangerous.
Most recently, Sikka co-authored an academic paper with his teenage son arguing mathematically that LLMs are fundamentally incapable of completing complex agentic tasks beyond a certain complexity threshold. A Stanford-trained PhD debating the limits of machine cognition with his own child. The apple, apparently, falls close.
Today Sikka holds board seats at Oracle, BMW Group, and GlaxoSmithKline - three radically different industries, each betting that his read on the technological future is worth having in the room. He also advises Stanford's Institute for Human-Centered AI. His wife, who calls him "a fake Punjabi" given his Gujarati upbringing, presumably has a more grounded view of the situation.