Budget 2026 must focus on making 'AI compute easy, predictable, and scalable', says SpeakX.ai CEO Arpit Mittal
"The government can play a massive role by becoming an early user of proven Indian models in public education and skilling programs," said Arpit Mittal, CEO of SpeakX.ai.
- Jan 27, 2026,
- Updated Jan 27, 2026 1:38 PM IST
As India is gearing up for the 2026 Union Budget, the conversation around artificial intelligence (AI) is becoming more practical and execution-focused. Arpit Mittal, CEO of SpeakX.ai, an Edtech startup, in an email interaction told Business Today that rather than bringing policies to strengthen AI, the government’s core focus should be rethinking funding mechanisms, infrastructure support and long-term enablement.
Founded in 2023 by Mittal, the Gurugram-based startup aims to bridge India's spoken English gap by combining AI with speech technology.
AI compute crunch
SpeakX.ai, in October 2025, raised $16 million (around Rs 142 crore) in funding led by WestBridge Capital. Hence, the company said that funding is not the biggest challenge anymore; the real problem is compute crunch.
“Startups need access from day one, not after months of paperwork, so upfront compute credits are far more useful than reimbursement models,” said Mittal.
He further expects the Budget 2026 to reduce import duties on GPUs and AI servers. Mittal added, “Affordable and reliable green power for data centres, along with faster approvals, would further bring down costs, since AI workloads consume large amounts of energy.”
He also emphasised that the IndiaAI Mission must offer transparent pricing and guaranteed uptime to support startups as they scale.
Milestone-based funding
Mittal told Business Today that the government should provide a disciplined "milestone-based" framework for deep-tech startups. “Instead of one-time grants, the government can help support startups through milestone-based funding linked to clear technical progress,” he said.
This will help Indian startups build AI models that are centered to solve "real Indian problems." He cited SpeakX.ai as an example, noting AI models should work well across “Indian languages, understanding local speech and accents, and supporting safe tutoring and assessment in education.”
In addition, he suggests that the government should be the early adopter of India-made AI models and integrate them into public education and skilling programs. “Real-world adoption at this level allows startups to test their solutions at scale, build trust and credibility, and grow into strong, sustainable businesses,” said Mittal.
Push for "Sovereign AI"
During the interaction, Mittal also highlighted the necessity for data security and stability for startups in India. He said that startups lose valuable months waiting for GPU access, since they have to rely on international providers. He also said that international providers limit long-term growth. Hence, “accelerating sovereign GPU clusters is critical,” said Mittal.
An India-owned cluster will not only save time, but it will bring stable pricing, local storage, and a predictable availability. It is also crucial for sectors like education and finance, and for Tier 2 and Tier 3 cities.
In addition, it will “serve the next 500 million Indians, startups need large-scale, reliable infrastructure within the country.”
India as a global AI hub
Mittal further talked about the government introducing a "plug-and-play" system in Budget 2026 to make AI computing resources easily accessible, reliable, and scalable. “When GPU access is affordable and reliable, founders can build, train, and deploy models without moving their teams or intellectual property abroad,” said Mittal.
“This is how India moves from being an AI consumer to becoming a global AI leader,” Mittal added in its conclusive remarks.
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine
As India is gearing up for the 2026 Union Budget, the conversation around artificial intelligence (AI) is becoming more practical and execution-focused. Arpit Mittal, CEO of SpeakX.ai, an Edtech startup, in an email interaction told Business Today that rather than bringing policies to strengthen AI, the government’s core focus should be rethinking funding mechanisms, infrastructure support and long-term enablement.
Founded in 2023 by Mittal, the Gurugram-based startup aims to bridge India's spoken English gap by combining AI with speech technology.
AI compute crunch
SpeakX.ai, in October 2025, raised $16 million (around Rs 142 crore) in funding led by WestBridge Capital. Hence, the company said that funding is not the biggest challenge anymore; the real problem is compute crunch.
“Startups need access from day one, not after months of paperwork, so upfront compute credits are far more useful than reimbursement models,” said Mittal.
He further expects the Budget 2026 to reduce import duties on GPUs and AI servers. Mittal added, “Affordable and reliable green power for data centres, along with faster approvals, would further bring down costs, since AI workloads consume large amounts of energy.”
He also emphasised that the IndiaAI Mission must offer transparent pricing and guaranteed uptime to support startups as they scale.
Milestone-based funding
Mittal told Business Today that the government should provide a disciplined "milestone-based" framework for deep-tech startups. “Instead of one-time grants, the government can help support startups through milestone-based funding linked to clear technical progress,” he said.
This will help Indian startups build AI models that are centered to solve "real Indian problems." He cited SpeakX.ai as an example, noting AI models should work well across “Indian languages, understanding local speech and accents, and supporting safe tutoring and assessment in education.”
In addition, he suggests that the government should be the early adopter of India-made AI models and integrate them into public education and skilling programs. “Real-world adoption at this level allows startups to test their solutions at scale, build trust and credibility, and grow into strong, sustainable businesses,” said Mittal.
Push for "Sovereign AI"
During the interaction, Mittal also highlighted the necessity for data security and stability for startups in India. He said that startups lose valuable months waiting for GPU access, since they have to rely on international providers. He also said that international providers limit long-term growth. Hence, “accelerating sovereign GPU clusters is critical,” said Mittal.
An India-owned cluster will not only save time, but it will bring stable pricing, local storage, and a predictable availability. It is also crucial for sectors like education and finance, and for Tier 2 and Tier 3 cities.
In addition, it will “serve the next 500 million Indians, startups need large-scale, reliable infrastructure within the country.”
India as a global AI hub
Mittal further talked about the government introducing a "plug-and-play" system in Budget 2026 to make AI computing resources easily accessible, reliable, and scalable. “When GPU access is affordable and reliable, founders can build, train, and deploy models without moving their teams or intellectual property abroad,” said Mittal.
“This is how India moves from being an AI consumer to becoming a global AI leader,” Mittal added in its conclusive remarks.
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine