1 hr ago
Gnani AI Pushes Enterprise Sovereign AI Beyond National Borders
Gnani AI builds artificial-intelligence models that understand speech and languages.
Its CEO says businesses want more control over the AI they use.
Companies worry that foreign AI providers might use their private data or turn off access to their models.
Gnani is developing a speech-to-speech model with help from the IndiaAI Mission.
The model is expected to be released in a few months, but no exact date was given.
Gnani also created Evon, a 30-billion-parameter text model focused on Indian languages.
The company is making an open-weight version so businesses can run and improve it on their own computers.
Gopalan says smaller models can be more useful when they are cheaper and trained for a company’s specific needs.
Gnani is seeking customers in India, West Asia, Japan, the Philippines and the United States.
Gnani AI CEO Ganesh Gopalan says sovereign AI now concerns enterprises as well as nations.
He says companies fear foreign AI providers could train on their data or control access to their models.
Gnani AI has raised $14.19 million in Series B funding and plans another external round for global expansion.
The company is developing a 14-billion-parameter speech-to-speech model through the IndiaAI Mission.
Gopalan says Gnani’s 30-billion-parameter Evon model prioritizes Indic languages, accuracy and cost over model size.
- Who
- Ganesh Gopalan, co-founder and CEO of Gnani AI, discussed the company’s enterprise AI strategy.
- What
- Gnani AI is developing sovereign and enterprise-focused AI models, including a 14-billion-parameter speech-to-speech model.
- Where
- The discussion occurred on the sidelines of the Global Fintech Fest in Mumbai.
- When
- The conversation took place on Friday; the speech-to-speech model is expected in a few months.
- Why
- Gopalan says enterprises want control over their data and AI systems while seeking better returns relative to computing costs.
Enterprise-Controlled AI
Foreign Frontier AI
Data sovereignty
Enterprise-Controlled AI
Gopalan argues that enterprises need models they can run and fine-tune themselves, protecting control over private data and business knowledge.
Foreign Frontier AI
Data residency can place data in a country while the underlying model remains controlled by a foreign provider, according to Gopalan’s concern.
Model size versus efficiency
Enterprise-Controlled AI
Gnani AI says smaller, specialized models can deliver better accuracy relative to computing cost for tasks such as payments, underwriting and claims.
Foreign Frontier AI
Larger frontier models remain part of the competing approach, although Gopalan argues that size alone does not determine enterprise value.
Open versus managed AI
Enterprise-Controlled AI
Gnani AI plans open-weight models that companies can download, operate on their own servers and fine-tune independently.
Foreign Frontier AI
The company will also offer closed premium versions and managed services for organizations that want outside help with customization.
Key facts
- Company
- Gnani AI
- Recent funding
- $14.19 million Series B
- Planned funding use
- Global go-to-market expansion
- Speech-to-speech model
- 14 billion parameters; launch expected in a few months
- Evon model
- 30 billion parameters and designed for Indic-language understanding and enterprise tasks
- Government support
- The IndiaAI Mission provided access to computing resources
- Target markets
- West Asia, Japan, the Philippines and the United States
Quotes
Ganesh Gopalan
Co-founder and CEO of Gnani AI
“With an open-weight model, companies can download it, run it on their own servers and fine-tune it on their own data, without even contacting us.”
financialexpress.com
“AI was once a shiny new object. Now enterprises want ROI.”
financialexpress.com








