Inside India’s scramble for AI independence

Stay Ahead, Stay ONMINE

Inside India’s scramble for AI independence

In Bengaluru, India, Adithya Kolavi felt a mix of excitement and validation as he watched DeepSeek unleash its disruptive language model on the world earlier this year. The Chinese technology rivaled the best of the West in terms of benchmarks, but it had been built with far less capital in far less time. “I thought: ‘This is how we disrupt with less,’” says Kolavi, the 20-year-old founder of the Indian AI startup CognitiveLab. “If DeepSeek could do it, why not us?” But for Abhishek Upperwal, founder of Soket AI Labs and architect of one of India’s earliest efforts to develop a foundation model, the moment felt more bittersweet. Upperwal’s model, called Pragna-1B, had struggled to stay afloat with tiny grants while he watched global peers raise millions. The multilingual model had a relatively modest 1.25 billion parameters and was designed to reduce the “language tax,” the extra costs that arise because India—unlike the US or even China—has a multitude of languages to support. His team had trained it, but limited resources meant it couldn’t scale. As a result, he says, the project became a proof of concept rather than a product. “If we had been funded two years ago, there’s a good chance we’d be the ones building what DeepSeek just released,” he says. Kolavi’s enthusiasm and Upperwal’s dismay reflect the spectrum of emotions among India’s AI builders. Despite its status as a global tech hub, the country lags far behind the likes of the US and China when it comes to homegrown AI. That gap has opened largely because India has chronically underinvested in R&D, institutions, and invention. Meanwhile, since no one native language is spoken by the majority of the population, training language models is far more complicated than it is elsewhere. Historically known as the global back office for the software industry, India has a tech ecosystem that evolved with a services-first mindset. Giants like Infosys and TCS built their success on efficient software delivery, but invention was neither prioritized nor rewarded. Meanwhile, India’s R&D spending hovered at just 0.65% of GDP ($25.4 billion) in 2024, far behind China’s 2.68% ($476.2 billion) and the US’s 3.5% ($962.3 billion). The muscle to invent and commercialize deep tech, from algorithms to chips, was just never built. Isolated pockets of world-class research do exist within government agencies like the DRDO (Defense Research & Development Organization) and ISRO (Indian Space Research Organization), but their breakthroughs rarely spill into civilian or commercial use. India lacks the bridges to connect risk-taking research to commercial pathways, the way DARPA does in the US. Meanwhile, much of India’s top talent migrates abroad, drawn to ecosystems that better understand and, crucially, fund deep tech.So when the open-source foundation model DeepSeek-R1 suddenly outperformed many global peers, it struck a nerve. This launch by a Chinese startup prompted Indian policymakers to confront just how far behind the country was in AI infrastructure, and how urgently it needed to respond. India responds In January 2025, 10 days after DeepSeek-R1’s launch, the Ministry of Electronics and Information Technology (MeitY) solicited proposals for India’s own foundation models, which are large AI models that can be adapted to a wide range of tasks. Its public tender invited private-sector cloud and data‑center companies to reserve GPU compute capacity for government‑led AI research. Providers including Jio, Yotta, E2E Networks, Tata, AWS partners, and CDAC responded. Through this arrangement, MeitY suddenly had access to nearly 19,000 GPUs at subsidized rates, repurposed from private infrastructure and allocated specifically to foundational AI projects. This triggered a surge of proposals from companies wanting to build their own models. Within two weeks, it had 67 proposals in hand. That number tripled by mid-March. In April, the government announced plans to develop six large-scale models by the end of 2025, plus 18 additional AI applications targeting sectors like agriculture, education, and climate action. Most notably, it tapped Sarvam AI to build a 70-billion-parameter model optimized for Indian languages and needs. For a nation long restricted by limited research infrastructure, things moved at record speed, marking a rare convergence of ambition, talent, and political will. “India could do a Mangalyaan in AI,” said Gautam Shroff of IIIT-Delhi, referencing the country’s cost-effective, and successful, Mars orbiter mission. Jaspreet Bindra, cofounder of AI&Beyond, an organization focused on teaching AI literacy, captured the urgency: “DeepSeek is probably the best thing that happened to India. It gave us a kick in the backside to stop talking and start doing something.” The language problem One of the most fundamental challenges in building foundational AI models for India is the country’s sheer linguistic diversity. With 22 official languages, hundreds of dialects, and millions of people who are multilingual, India poses a problem that few existing LLMs are equipped to handle. Whereas a massive amount of high-quality web data is available in English, Indian languages collectively make up less than 1% of online content. The lack of digitized, labeled, and cleaned data in languages like Bhojpuri and Kannada makes it difficult to train LLMs that understand how Indians actually speak or search. Global tokenizers, which break text into units a model can process, also perform poorly on many Indian scripts, misinterpreting characters or skipping some altogether. As a result, even when Indian languages are included in multilingual models, they’re often poorly understood and inaccurately generated. And unlike OpenAI and DeepSeek, which achieved scale using structured English-language data, Indian teams often begin with fragmented and low-quality data sets encompassing dozens of Indian languages. This makes the early steps of training foundation models far more complex. Nonetheless, a small but determined group of Indian builders is starting to shape the country’s AI future. For example, Sarvam AI has created OpenHathi-Hi-v0.1, an open-source Hindi language model that shows the Indian AI field’s growing ability to address the country’s vast linguistic diversity. The model, built on Meta’s Llama 2 architecture, was trained on 40 billion tokens of Hindi and related Indian-language content, making it one of the largest open-source Hindi models available to date. Pragna-1B, the multilingual model from Upperwal, is more evidence that India could solve for its own linguistic complexity. Trained on 300 billion tokens for just $250,000, it introduced a technique called “balanced tokenization” to address a unique challenge in Indian AI, enabling a 1.25-billion-parameter model to behave like a much larger one.The issue is that Indian languages use complex scripts and agglutinative grammar, where words are formed by stringing together many smaller units of meaning using prefixes and suffixes. Unlike English, which separates words with spaces and follows relatively simple structures, Indian languages like Hindi, Tamil, and Kannada often lack clear word boundaries and pack a lot of information into single words. Standard tokenizers struggle with such inputs. They end up breaking Indian words into too many tokens, which bloats the input and makes it harder for models to understand the meaning efficiently or respond accurately. With the new technique, however, “a billion-parameter model was equivalent to a 7 billion one like Llama 2,” Upperwal says. This performance was particularly marked in Hindi and Gujarati, where global models often underperform because of limited multilingual training data. It was a reminder that with smart engineering, small teams could still push boundaries.Upperwal eventually repurposed his core tech to build speech APIs for 22 Indian languages, a more immediate solution better suited to rural users who are often left out of English-first AI experiences. “If the path to AGI is a hundred-step process, training a language model is just step one,” he says. At the other end of the spectrum are startups with more audacious aims. Krutrim-2, for instance, is a 12-billion-parameter multilingual language model optimized for English and 22 Indian languages. Krutrim-2 is attempting to solve India’s specific problems of linguistic diversity, low-quality data, and cost constraints. The team built a custom Indic tokenizer, optimized training infrastructure, and designed models for multimodal and voice-first use cases from the start, crucial in a country where text interfaces can be a problem. Krutrim’s bet is that its approach will not only enable Indian AI sovereignty but also offer a model for AI that works across the Global South. Besides public funding and compute infrastructure, India also needs the institutional support of talent, the research depth, and the long-horizon capital that produce globally competitive science. While venture capital still hesitates to bet on research, new experiments are emerging. Paras Chopra, an entrepreneur who previously built and sold the software-as-a-service company Wingify, is now personally funding Lossfunk, a Bell Labs–style AI residency program designed to attract independent researchers with a taste for open-source science. “We don’t have role models in academia or industry,” says Chopra. “So we’re creating a space where top researchers can learn from each other and have startup-style equity upside.” Government-backed bet on sovereign AI The clearest marker of India’s AI ambitions came when the government selected Sarvam AI to develop a model focused on Indian languages and voice fluency. The idea is that it would not only help Indian companies compete in the global AI arms race but benefit the wider population as well. “If it becomes part of the India stack, you can educate hundreds of millions through conversational interfaces,” says Bindra. Sarvam was given access to 4,096 Nvidia H100 GPUs for training a 70-billion-parameter Indian language model over six months. (The company previously released a 2-billion-parameter model trained in 10 Indian languages, called Sarvam-1.) Sarvam’s project and others are part of a larger strategy called the IndiaAI Mission, a $1.25 billion national initiative launched in March 2024 to build out India’s core AI infrastructure and make advanced tools more widely accessible. Led by MeitY, the mission is focused on supporting AI startups, particularly those developing foundation models in Indian languages and applying AI to key sectors such as health care, education, and agriculture. Under its compute program, the government is deploying more than 18,000 GPUs, including nearly 13,000 high-end H100 chips, to a select group of Indian startups that currently includes Sarvam, Upperwal’s Soket Labs, Gnani AI, and Gan AI. The mission also includes plans to launch a national multilingual data set repository, establish AI labs in smaller cities, and fund deep-tech R&D. The broader goal is to equip Indian developers with the infrastructure needed to build globally competitive AI and ensure that the results are grounded in the linguistic and cultural realities of India and the Global South.According to Abhishek Singh, CEO of IndiaAI and an officer with MeitY, India’s broader push into deep tech is expected to raise around $12 billion in research and development investment over the next five years. This includes approximately $162 million through the IndiaAI Mission, with about $32 million earmarked for direct startup funding. The National Quantum Mission is contributing another $730 million to support India’s ambitions in quantum research. In addition to this, the national budget document for 2025-26 announced a $1.2 billion Deep Tech Fund of Funds aimed at catalyzing early-stage innovation in the private sector. The rest, nearly $9.9 billion, is expected to come from private and international sources including corporate R&D, venture capital firms, high-net-worth individuals, philanthropists, and global technology leaders such as Microsoft. IndiaAI has now received more than 500 applications from startups proposing use cases in sectors like health, governance, and agriculture. “We’ve already announced support for Sarvam, and 10 to 12 more startups will be funded solely for foundational models,” says Singh. Selection criteria include access to training data, talent depth, sector fit, and scalability. Open or closed? The IndiaAI program, however, is not without controversy. Sarvam is being built as a closed model, not open-source, despite its public tech roots. That has sparked debate about the proper balance between private enterprise and the public good. “True sovereignty should be rooted in openness and transparency,” says Amlan Mohanty, an AI policy specialist. He points to DeepSeek-R1, which despite its 236-billion parameter size was made freely available for commercial use. Its release allowed developers around the world to fine-tune it on low-cost GPUs, creating faster variants and extending its capabilities to non-English applications. “Releasing an open-weight model with efficient inference can democratize AI,” says Hancheng Cao, an assistant professor of information systems and operations management at Emory University. “It makes it usable by developers who don’t have massive infrastructure.” IndiaAI, however, has taken a neutral stance on whether publicly funded models should be open-source. “We didn’t want to dictate business models,” says Singh. “India has always supported open standards and open source, but it’s up to the teams. The goal is strong Indian models, whatever the route.” There are other challenges as well. In late May, Sarvam AI unveiled Sarvam‑M, a 24-billion-parameter multilingual LLM fine-tuned for 10 Indian languages and built on top of Mistral Small, an efficient model developed by the French company Mistral AI. Sarvam’s cofounder Vivek Raghavan called the model “an important stepping stone on our journey to build sovereign AI for India.” But its download numbers were underwhelming, with only 300 in the first two days. The venture capitalist Deedy Das called the launch “embarrassing.”And the issues go beyond the lukewarm early reception. Many developers in India still lack easy access to GPUs and the broader ecosystem for Indian-language AI applications is still nascent. The compute question Compute scarcity is emerging as one of the most significant bottlenecks in generative AI, not just in India but across the globe. For countries still heavily reliant on imported GPUs and lacking domestic fabrication capacity, the cost of building and running large models is often prohibitive. India still imports most of its chips rather than producing them domestically, and training large models remains expensive. That’s why startups and researchers alike are focusing on software-level efficiencies that involve smaller models, better inference, and fine-tuning frameworks that optimize for performance on fewer GPUs. “The absence of infrastructure doesn’t mean the absence of innovation,” says Cao. “Supporting optimization science is a smart way to work within constraints.” Yet Singh of IndiaAI argues that the tide is turning on the infrastructure challenge thanks to the new government programs and private-public partnerships. “I believe that within the next three months, we will no longer face the kind of compute bottlenecks we saw last year,” he says. India also has a cost advantage.According to Gupta, building a hyperscale data center in India costs about $5 million, roughly half what it would cost in markets like the US, Europe, or Singapore. That’s thanks to affordable land, lower construction and labor costs, and a large pool of skilled engineers. For now, India’s AI ambitions seem less about leapfrogging OpenAI or DeepSeek and more about strategic self-determination. Whether its approach takes the form of smaller sovereign models, open ecosystems, or public-private hybrids, the country is betting that it can chart its own course. While some experts argue that the government’s action, or reaction (to DeepSeek), is performative and aligned with its nationalistic agenda, many startup founders are energized. They see the growing collaboration between the state and the private sector as a real opportunity to overcome India’s long-standing structural challenges in tech innovation. At a Meta summit held in Bengaluru last year, Nandan Nilekani, the chairman of Infosys, urged India to resist chasing a me-too AI dream. “Let the big boys in the Valley do it,” he said of building LLMs. “We will use it to create synthetic data, build small language models quickly, and train them using appropriate data.” His view that India should prioritize strength over spectacle had a divided reception. But it reflects a broader growing consensus on whether India should play a different game altogether. “Trying to dominate every layer of the stack isn’t realistic, even for China,” says Bharath Reddy, a researcher at the Takshashila Institution, an Indian public policy nonprofit. “Dominate one layer, like applications, services, or talent, so you remain indispensable.”

“I thought: ‘This is how we disrupt with less,’” says Kolavi, the 20-year-old founder of the Indian AI startup CognitiveLab. “If DeepSeek could do it, why not us?”

But for Abhishek Upperwal, founder of Soket AI Labs and architect of one of India’s earliest efforts to develop a foundation model, the moment felt more bittersweet.

Upperwal’s model, called Pragna-1B, had struggled to stay afloat with tiny grants while he watched global peers raise millions. The multilingual model had a relatively modest 1.25 billion parameters and was designed to reduce the “language tax,” the extra costs that arise because India—unlike the US or even China—has a multitude of languages to support. His team had trained it, but limited resources meant it couldn’t scale. As a result, he says, the project became a proof of concept rather than a product.

“If we had been funded two years ago, there’s a good chance we’d be the ones building what DeepSeek just released,” he says.

Kolavi’s enthusiasm and Upperwal’s dismay reflect the spectrum of emotions among India’s AI builders. Despite its status as a global tech hub, the country lags far behind the likes of the US and China when it comes to homegrown AI. That gap has opened largely because India has chronically underinvested in R&D, institutions, and invention. Meanwhile, since no one native language is spoken by the majority of the population, training language models is far more complicated than it is elsewhere.

Historically known as the global back office for the software industry, India has a tech ecosystem that evolved with a services-first mindset. Giants like Infosys and TCS built their success on efficient software delivery, but invention was neither prioritized nor rewarded. Meanwhile, India’s R&D spending hovered at just 0.65% of GDP ($25.4 billion) in 2024, far behind China’s 2.68% ($476.2 billion) and the US’s 3.5% ($962.3 billion). The muscle to invent and commercialize deep tech, from algorithms to chips, was just never built.

Isolated pockets of world-class research do exist within government agencies like the DRDO (Defense Research & Development Organization) and ISRO (Indian Space Research Organization), but their breakthroughs rarely spill into civilian or commercial use. India lacks the bridges to connect risk-taking research to commercial pathways, the way DARPA does in the US. Meanwhile, much of India’s top talent migrates abroad, drawn to ecosystems that better understand and, crucially, fund deep tech.

So when the open-source foundation model DeepSeek-R1 suddenly outperformed many global peers, it struck a nerve. This launch by a Chinese startup prompted Indian policymakers to confront just how far behind the country was in AI infrastructure, and how urgently it needed to respond.

India responds

In January 2025, 10 days after DeepSeek-R1’s launch, the Ministry of Electronics and Information Technology (MeitY) solicited proposals for India’s own foundation models, which are large AI models that can be adapted to a wide range of tasks. Its public tender invited private-sector cloud and data‑center companies to reserve GPU compute capacity for government‑led AI research.

Providers including Jio, Yotta, E2E Networks, Tata, AWS partners, and CDAC responded. Through this arrangement, MeitY suddenly had access to nearly 19,000 GPUs at subsidized rates, repurposed from private infrastructure and allocated specifically to foundational AI projects. This triggered a surge of proposals from companies wanting to build their own models.

Within two weeks, it had 67 proposals in hand. That number tripled by mid-March.

In April, the government announced plans to develop six large-scale models by the end of 2025, plus 18 additional AI applications targeting sectors like agriculture, education, and climate action. Most notably, it tapped Sarvam AI to build a 70-billion-parameter model optimized for Indian languages and needs.

For a nation long restricted by limited research infrastructure, things moved at record speed, marking a rare convergence of ambition, talent, and political will.

“India could do a Mangalyaan in AI,” said Gautam Shroff of IIIT-Delhi, referencing the country’s cost-effective, and successful, Mars orbiter mission.

Jaspreet Bindra, cofounder of AI&Beyond, an organization focused on teaching AI literacy, captured the urgency: “DeepSeek is probably the best thing that happened to India. It gave us a kick in the backside to stop talking and start doing something.”

The language problem

One of the most fundamental challenges in building foundational AI models for India is the country’s sheer linguistic diversity. With 22 official languages, hundreds of dialects, and millions of people who are multilingual, India poses a problem that few existing LLMs are equipped to handle.

Whereas a massive amount of high-quality web data is available in English, Indian languages collectively make up less than 1% of online content. The lack of digitized, labeled, and cleaned data in languages like Bhojpuri and Kannada makes it difficult to train LLMs that understand how Indians actually speak or search.

Global tokenizers, which break text into units a model can process, also perform poorly on many Indian scripts, misinterpreting characters or skipping some altogether. As a result, even when Indian languages are included in multilingual models, they’re often poorly understood and inaccurately generated.

And unlike OpenAI and DeepSeek, which achieved scale using structured English-language data, Indian teams often begin with fragmented and low-quality data sets encompassing dozens of Indian languages. This makes the early steps of training foundation models far more complex.

Nonetheless, a small but determined group of Indian builders is starting to shape the country’s AI future.

For example, Sarvam AI has created OpenHathi-Hi-v0.1, an open-source Hindi language model that shows the Indian AI field’s growing ability to address the country’s vast linguistic diversity. The model, built on Meta’s Llama 2 architecture, was trained on 40 billion tokens of Hindi and related Indian-language content, making it one of the largest open-source Hindi models available to date.

Pragna-1B, the multilingual model from Upperwal, is more evidence that India could solve for its own linguistic complexity. Trained on 300 billion tokens for just $250,000, it introduced a technique called “balanced tokenization” to address a unique challenge in Indian AI, enabling a 1.25-billion-parameter model to behave like a much larger one.

The issue is that Indian languages use complex scripts and agglutinative grammar, where words are formed by stringing together many smaller units of meaning using prefixes and suffixes. Unlike English, which separates words with spaces and follows relatively simple structures, Indian languages like Hindi, Tamil, and Kannada often lack clear word boundaries and pack a lot of information into single words. Standard tokenizers struggle with such inputs. They end up breaking Indian words into too many tokens, which bloats the input and makes it harder for models to understand the meaning efficiently or respond accurately.

With the new technique, however, “a billion-parameter model was equivalent to a 7 billion one like Llama 2,” Upperwal says. This performance was particularly marked in Hindi and Gujarati, where global models often underperform because of limited multilingual training data. It was a reminder that with smart engineering, small teams could still push boundaries.

Upperwal eventually repurposed his core tech to build speech APIs for 22 Indian languages, a more immediate solution better suited to rural users who are often left out of English-first AI experiences.

“If the path to AGI is a hundred-step process, training a language model is just step one,” he says.

At the other end of the spectrum are startups with more audacious aims. Krutrim-2, for instance, is a 12-billion-parameter multilingual language model optimized for English and 22 Indian languages.

Krutrim-2 is attempting to solve India’s specific problems of linguistic diversity, low-quality data, and cost constraints. The team built a custom Indic tokenizer, optimized training infrastructure, and designed models for multimodal and voice-first use cases from the start, crucial in a country where text interfaces can be a problem.

Krutrim’s bet is that its approach will not only enable Indian AI sovereignty but also offer a model for AI that works across the Global South.

Besides public funding and compute infrastructure, India also needs the institutional support of talent, the research depth, and the long-horizon capital that produce globally competitive science.

While venture capital still hesitates to bet on research, new experiments are emerging. Paras Chopra, an entrepreneur who previously built and sold the software-as-a-service company Wingify, is now personally funding Lossfunk, a Bell Labs–style AI residency program designed to attract independent researchers with a taste for open-source science.

“We don’t have role models in academia or industry,” says Chopra. “So we’re creating a space where top researchers can learn from each other and have startup-style equity upside.”

Government-backed bet on sovereign AI

The clearest marker of India’s AI ambitions came when the government selected Sarvam AI to develop a model focused on Indian languages and voice fluency.

The idea is that it would not only help Indian companies compete in the global AI arms race but benefit the wider population as well. “If it becomes part of the India stack, you can educate hundreds of millions through conversational interfaces,” says Bindra.

Sarvam was given access to 4,096 Nvidia H100 GPUs for training a 70-billion-parameter Indian language model over six months. (The company previously released a 2-billion-parameter model trained in 10 Indian languages, called Sarvam-1.)

Sarvam’s project and others are part of a larger strategy called the IndiaAI Mission, a $1.25 billion national initiative launched in March 2024 to build out India’s core AI infrastructure and make advanced tools more widely accessible. Led by MeitY, the mission is focused on supporting AI startups, particularly those developing foundation models in Indian languages and applying AI to key sectors such as health care, education, and agriculture.

Under its compute program, the government is deploying more than 18,000 GPUs, including nearly 13,000 high-end H100 chips, to a select group of Indian startups that currently includes Sarvam, Upperwal’s Soket Labs, Gnani AI, and Gan AI.

The mission also includes plans to launch a national multilingual data set repository, establish AI labs in smaller cities, and fund deep-tech R&D. The broader goal is to equip Indian developers with the infrastructure needed to build globally competitive AI and ensure that the results are grounded in the linguistic and cultural realities of India and the Global South.

According to Abhishek Singh, CEO of IndiaAI and an officer with MeitY, India’s broader push into deep tech is expected to raise around $12 billion in research and development investment over the next five years.

This includes approximately $162 million through the IndiaAI Mission, with about $32 million earmarked for direct startup funding. The National Quantum Mission is contributing another $730 million to support India’s ambitions in quantum research. In addition to this, the national budget document for 2025-26 announced a $1.2 billion Deep Tech Fund of Funds aimed at catalyzing early-stage innovation in the private sector.

The rest, nearly $9.9 billion, is expected to come from private and international sources including corporate R&D, venture capital firms, high-net-worth individuals, philanthropists, and global technology leaders such as Microsoft.

IndiaAI has now received more than 500 applications from startups proposing use cases in sectors like health, governance, and agriculture.

“We’ve already announced support for Sarvam, and 10 to 12 more startups will be funded solely for foundational models,” says Singh. Selection criteria include access to training data, talent depth, sector fit, and scalability.

Open or closed?

The IndiaAI program, however, is not without controversy. Sarvam is being built as a closed model, not open-source, despite its public tech roots. That has sparked debate about the proper balance between private enterprise and the public good.

“True sovereignty should be rooted in openness and transparency,” says Amlan Mohanty, an AI policy specialist. He points to DeepSeek-R1, which despite its 236-billion parameter size was made freely available for commercial use.

Its release allowed developers around the world to fine-tune it on low-cost GPUs, creating faster variants and extending its capabilities to non-English applications.

“Releasing an open-weight model with efficient inference can democratize AI,” says Hancheng Cao, an assistant professor of information systems and operations management at Emory University. “It makes it usable by developers who don’t have massive infrastructure.”

IndiaAI, however, has taken a neutral stance on whether publicly funded models should be open-source.

“We didn’t want to dictate business models,” says Singh. “India has always supported open standards and open source, but it’s up to the teams. The goal is strong Indian models, whatever the route.”

There are other challenges as well. In late May, Sarvam AI unveiled Sarvam‑M, a 24-billion-parameter multilingual LLM fine-tuned for 10 Indian languages and built on top of Mistral Small, an efficient model developed by the French company Mistral AI. Sarvam’s cofounder Vivek Raghavan called the model “an important stepping stone on our journey to build sovereign AI for India.” But its download numbers were underwhelming, with only 300 in the first two days. The venture capitalist Deedy Das called the launch “embarrassing.”

And the issues go beyond the lukewarm early reception. Many developers in India still lack easy access to GPUs and the broader ecosystem for Indian-language AI applications is still nascent.

The compute question

Compute scarcity is emerging as one of the most significant bottlenecks in generative AI, not just in India but across the globe. For countries still heavily reliant on imported GPUs and lacking domestic fabrication capacity, the cost of building and running large models is often prohibitive.

India still imports most of its chips rather than producing them domestically, and training large models remains expensive. That’s why startups and researchers alike are focusing on software-level efficiencies that involve smaller models, better inference, and fine-tuning frameworks that optimize for performance on fewer GPUs.

“The absence of infrastructure doesn’t mean the absence of innovation,” says Cao. “Supporting optimization science is a smart way to work within constraints.”

Yet Singh of IndiaAI argues that the tide is turning on the infrastructure challenge thanks to the new government programs and private-public partnerships. “I believe that within the next three months, we will no longer face the kind of compute bottlenecks we saw last year,” he says.

India also has a cost advantage.

According to Gupta, building a hyperscale data center in India costs about $5 million, roughly half what it would cost in markets like the US, Europe, or Singapore. That’s thanks to affordable land, lower construction and labor costs, and a large pool of skilled engineers.

For now, India’s AI ambitions seem less about leapfrogging OpenAI or DeepSeek and more about strategic self-determination. Whether its approach takes the form of smaller sovereign models, open ecosystems, or public-private hybrids, the country is betting that it can chart its own course.

While some experts argue that the government’s action, or reaction (to DeepSeek), is performative and aligned with its nationalistic agenda, many startup founders are energized. They see the growing collaboration between the state and the private sector as a real opportunity to overcome India’s long-standing structural challenges in tech innovation.

At a Meta summit held in Bengaluru last year, Nandan Nilekani, the chairman of Infosys, urged India to resist chasing a me-too AI dream.

“Let the big boys in the Valley do it,” he said of building LLMs. “We will use it to create synthetic data, build small language models quickly, and train them using appropriate data.”

His view that India should prioritize strength over spectacle had a divided reception. But it reflects a broader growing consensus on whether India should play a different game altogether.

“Trying to dominate every layer of the stack isn’t realistic, even for China,” says Bharath Reddy, a researcher at the Takshashila Institution, an Indian public policy nonprofit. “Dominate one layer, like applications, services, or talent, so you remain indispensable.”

Stay Ahead

Explore More Insights

Stay ahead with more perspectives on cutting-edge power, infrastructure, energy, bitcoin and AI solutions. Explore these articles to uncover strategies and insights shaping the future of industries.

A timeline of Broadcom/VMware and Siemens licensing dispute

June 24: VMware responds, saying that Siemens distributed infringing VMware products to its US subsidiaries in violation of US copyright law by accessing VMware’s US server. July 1: Nah uh, says Siemens. First, any actions taken by the parent company occurred in Germany. Also, downloading allegedly copyrighted software does not

JPMorgan launches carbon market blockchain app

Dive Brief: JPMorgan Chase is working to allow voluntary carbon markets to issue blockchain tokens at the registry level that represent ownership of carbon credits, permitting market participants to issue, transfer and retire credits, the bank announced Wednesday. JPMorgan is currently exploring testing processes with carbon registries from S&P Global

IBM Power11 challenges x86 and GPU giants with security-first server strategy

The IBM Power Cyber Vault solution is designed to provide protection against cyberattacks such as data corruption and encryption with proactive immutable snapshots that are automatically captured, stored, and tested on a custom-defined schedule, IBM said. Power11 also uses NIST-approved built-in quantum-safe cryptography designed to help protect systems from harvest-now, decrypt-later attacks

With Power 11, IBM bets big on security and reliability instead of taking Nvidia head-on

Working around the competition The launch invites comparisons with other AI chips on the market, but analysts say IBM’s Power11 servers are built for targeted use in specific industries. “These systems don’t compete on raw performance with offerings from Intel or Nvidia, but IBM can offer a stronger value proposition

Energy Department Authorizes Strategic Petroleum Reserve Exchange to Support Fuel Supply in Gulf Coast

WASHINGTON—The U.S. Department of Energy (DOE) today announced the authorization of an exchange from the Strategic Petroleum Reserve (SPR) with ExxonMobil Corporation to address logistical challenges impacting crude oil deliveries to the company’s Baton Rouge refinery. U.S. Secretary of Energy Chris Wright authorized this action to help maintain stable regional supply of transportation fuels across Louisiana and the broader Gulf Coast. This action preserves the SPR’s operational flexibility and will not impact or delay the Department’s ongoing efforts to refill the reserve. Under the exchange agreement, DOE will provide up to 1 million barrels of crude oil from the SPR. The exchange will support ExxonMobil’s restoration of refinery operations that were reduced due to an offshore supply disruption. ExxonMobil will return the borrowed crude along with additional barrels of crude oil for the SPR at no cost to the taxpayer. The Department remains in close coordination with industry partners to ensure stability in the fuel supply chain during the peak demand season. DOE continues to encourage refiners to prioritize efficient production and delivery of refined fuels, stands ready to support the nation’s energy security through the responsible use of strategic resources, and will continue to deliver on President Trump’s commitment to protect American energy security by refilling the SPR. Background: Sections 159 and 160 of the Energy Policy and Conservation Act (EPCA), 42 U.S.C.A. §§ 6239 and 6240, authorize the Secretary of Energy to exchange SPR petroleum products and to acquire petroleum products by exchange for storage in the SPR. The Secretary of Energy has previously exercised this legal authority to conduct emergency exchanges in response to supply disruptions, including Keystone Pipeline in 2022, and the Calcasieu Ship Channel closures in 2006 and 2000. An oil supply disruption has led to reduced operations at the Baton Rouge refinery, limiting production

Misaligned interconnection, transmission planning could hurt competitive markets: FERC’s Chang

There is a “misalignment” in grid interconnection and transmission planning processes that could harm competitive power markets, according to Judy Chang, a member of the Federal Energy Regulatory Commission. Amid surging electric demand forecasts, the interconnection logjam has led grid operators to propose short-term solutions, Chang said Thursday at a meeting held in Woodstock, Vermont, by WIRES, a transmission-focused trade group. The PJM Interconnection, the Midcontinent Independent System Operator and the Southwest Power Pool have proposed one-time processes that would create a fast-track interconnection review for planned generating projects that meet certain criteria. FERC approved PJM’s plan earlier this year over the opposition of some renewable energy companies that contend selected projects will be able to unfairly jump ahead of others that have been waiting in interconnection queues. “I don’t really love short-term fixes,” Chang said. “I really prefer to have better processes — fair and competitive processes — so that generators interconnecting know the rules of the game.” Chang said various issues are coming to a head at the same time, including disputes over interconnection cost allocation and the system’s ability to upgrade interconnection infrastructure and bring on new generation as fast as possible. Those issues could affect the future of competitive power markets, Chang said, noting that some states are considering withdrawing from regional transmission organizations. “I worry about how much states might want to compromise … the open access and competitive access to transmission and competitive markets by pulling back and finding internal solutions, or by complaining about competitive markets not meeting the challenge of the day,” she said. Getting generation online as quickly as possible is a key priority, according to Chang. “We should plan, design, permit — all faster,” Chang said. “So I am a big supporter of permitting reform on all infrastructure, but also

Environmental, consumer advocates sue Bonneville for joining SPP’s day-ahead market

Five advocacy groups on Thursday sued the Bonneville Power Administration for deciding to join the Southwest Power Pool’s Markets+ day-ahead market. The BPA’s May 9 decision to join Markets+ instead of a day-ahead market for the West developed by the California Independent System Operator will lead to higher electric costs, inefficient operations between market seams and potentially the need to build more generating facilities, according to the lawsuit filed by NW Energy Coalition, the Idaho Conservation League, the Montana Environmental Information Center, the Oregon Citizens’ Utility Board and the Sierra Club. “Bonneville’s decision on markets will affect the transmission and generation of electric power across the West and is exactly the type of major federal action that should first consider the harms it could cause to our air quality, grid system reliability, [and] fish and wildlife,” Jaimini Parekh, a senior attorney for Earthjustice, said in a press release. Earthjustice represents the advocacy groups. The BPA’s decision violated the Pacific Northwest Electric Power Planning and Conservation Act, the National Environmental Policy Act and the Administrative Procedure Act, according to the groups. “BPA did not rationally explain how joining the smaller, non-contiguous Markets+ footprint will enable it to meet its duty to promote an adequate, efficient, economical, and reliable power supply for the region that also gives priority to clean, renewable resources,” the groups said in the lawsuit filed at the U.S. Court of Appeals for the Ninth Circuit. The groups also contend that BPA’s choice to join Markets+ will likely increase the risk of blackouts during periods of high or extreme electricity demand because of the “many and complex” seams that power must be transferred across in the market as compared to CAISO’s Extended Day-Ahead Market or a “no-action” alternative. BPA does not comment on active litigation, said Nick Quinata, a

Ignoring customers is unsustainable and bad policy

Kent Chandler is a resident senior fellow in energy and environmental policy at the R Street Institute, Chris Villarreal is an associate fellow in energy and environmental policy at R Street and Michael Giberson is a senior fellow in energy policy at R Street. Electric bills are rising, and energy adequacy is at the top of policymakers’ agendas. The most recent North American Electric Reliability Corp. analysis indicates that much of the country is at an elevated risk this summer of experiencing insufficient operating reserves in above-normal conditions. Given the backlogs of generator interconnection requests and gas turbine orders, the delays inherent in current state and federal siting regimes, and the immediacy of load growth in certain parts of the country, it is clear we can’t build our way out of this situation in the near-term. However, enabling flexible customer demand can drive significant system-wide benefits, mitigating the cost or timing of expensive system upgrades. Now is the time for states to take action to empower consumers to play a greater role in their individual and collective energy future. It’s been 30 years since some states began restructuring their electricity laws, allowing customers to buy electricity from suppliers other than their incumbent utilities. However, even with significant leaps in technological innovation and expansion of wholesale markets, implementation of retail choice and customer empowerment within vertically-integrated states has lagged. To determine the state of play for customers to make their own electricity choices, we recently completed a review of consumer choice, retail competition and customer empowerment in all 50 states and the District of Columbia, culminating in a report grading the states on their level of customer empowerment. As you might imagine, the grades vary, from Texas receiving an A- to Alabama receiving the effort’s lone F. Regardless of the grade

XCF Targets $1B Global SAF Production Portfolio

XCF Global Inc. announced Thursday an investment plan of nearly $1 billion over the next three years to build sustainable aviation fuel (SAF) production facilities, with most of its in-the-pipeline projects in the United States. The package includes $350 million already invested in New Rise Reno in Nevada, which has a nameplate capacity of 38 million gallons a year. It started production early this year. The announcement follows XCF Global Capital Inc. and Focus Impact BH3 Acquisition Co.’s completion of their merger, creating what they said is the first publicly traded pure-play SAF producer in the U.S. Under its U.S. expansion plan, XCF has acquired three sites “ready for development”, it said in an online statement Thursday. The projects are planned to each have a nameplate capacity of 40 million gallons per annum. XCF expects to put them into operation by 2028. Expected to be completed 2027, New Rise Reno 2 will rise next to the existing plant, “enabling economies of scale and leveraging shared utilities and logistics infrastructure”, XCF said. Another project in Ft. Myers, Florida, will be built on a site with access to port infrastructure. It is expected to be completed 2028. The third, also targeted for start-up 2028, will rise in Wilson, North Carolina, eyeing East Coast markets. “These new sites are expected to replicate New Rise Reno’s modular, patent-pending site design and bundled technology stack, allowing for rapid deployment, flexible production, and capital-efficient scaling”, XCF said. “Each facility is expected to have the ability to produce multiple renewable fuel products, including SAF and renewable diesel, supporting a multi-product revenue strategy that maximizes plant utilization and financial performance”. XCF is also pursuing other “high-potential” markets, the statement said. Last month it announced a non-binding memorandum of understanding with Continual Renewable Ventures for an Australian SAF and

ADNOC Bags 3-Year LNG Order from SEFE

ADNOC Gas PLC will supply Germany’s state-owned SEFE Securing Energy for Europe GmbH 700,000 metric tons a year of liquefied natural gas (LNG) for three years starting 2025, the companies said Thursday. Abu Dhabi National Oil Co.’s gas processing and sales arm will source the LNG from the Das Island liquefaction facility, which has a capacity of six million metric tons per annum (MMtpa). The contract is valued about $400 million. “Das Island’s LNG plant has shipped over 3,500 LNG cargoes worldwide since starting operations in 1977, strengthening ADNOC Gas’ long-term relationships with key global energy partners”, a joint statement said. SEFE chief commercial officer Frederic Barnaud commented, “This new medium-term LNG contract builds on the long-term supply agreement with ADNOC that we signed last year, thereby adding another flexible source of LNG to our portfolio – to the benefit of both Europe’s security of supply and our global market trading activities”. In November 2024, ADNOC and SEFE announced an agreement under which ADNOC will deliver one MMtpa of LNG from the Ruwais LNG project to SEFE for 15 years. Targeted to start production 2028, the 9.6-MMtpa facility on the Persian Gulf coast would more than double ADNOC’s LNG output. ADNOC announced a positive final investment decision June 2024. SEFE-Venture Global Deal Separately on Wednesday SEFE and Venture Global Inc. said they had finalized an agreement to increase SEFE’s offtake from the CP2 LNG project in Cameron Parish, Louisiana, to three MMtpa. The deal adds 750,000 metric tons a year to their 2023 agreement. “Venture Global is expected to become Germany’s largest LNG supplier, with a combined 5 MTPA [million metric tons per annum] of 20-year offtake agreements signed with SEFE and EnBW”, the Arlington, Virginia-based developer said in a statement online. “In addition to its existing long-term agreements, Venture Global

Nvidia hits $4T market cap as AI, high-performance semiconductors hit stride

“The company added $1 trillion in market value in less than a year, a pace that surpasses Apple and Microsoft’s previous trajectories. This rapid ascent reflects how indispensable AI chipmakers have become in today’s digital economy,” Kiran Raj, practice head, Strategic Intelligence (Disruptor) at GlobalData, said in a statement. According to GlobalData’s Innovation Radar report, “AI Chips – Trends, Market Dynamics and Innovations,” the global AI chip market is projected to reach $154 billion by 2030, growing at a compound annual growth rate (CAGR) of 20%. Nvidia has much of that market, but it also has a giant bullseye on its back with many competitors gunning for its crown. “With its AI chips powering everything from data centers and cloud computing to autonomous vehicles and robotics, Nvidia is uniquely positioned. However, competitive pressure is mounting. Players like AMD, Intel, Google, and Huawei are doubling down on custom silicon, while regulatory headwinds and export restrictions are reshaping the competitive dynamics,” he said.

Enterprises will strengthen networks to take on AI, survey finds

Private data centers: 29.5% Traditional public cloud: 35.4% GPU as a service specialists: 18.5% Edge compute: 16.6% “There is little variation from training to inference, but the general pattern is workloads are concentrated a bit in traditional public cloud and then hyperscalers have significant presence in private data centers,” McGillicuddy explained. “There is emerging interest around deploying AI workloads at the corporate edge and edge compute environments as well, which allows them to have workloads residing closer to edge data in the enterprise, which helps them combat latency issues and things like that. The big key takeaway here is that the typical enterprise is going to need to make sure that its data center network is ready to support AI workloads.” AI networking challenges The popularity of AI doesn’t remove some of the business and technical concerns that the technology brings to enterprise leaders. According to the EMA survey, business concerns include security risk (39%), cost/budget (33%), rapid technology evolution (33%), and networking team skills gaps (29%). Respondents also indicated several concerns around both data center networking issues and WAN issues. Concerns related to data center networking included: Integration between AI network and legacy networks: 43% Bandwidth demand: 41% Coordinating traffic flows of synchronized AI workloads: 38% Latency: 36% WAN issues respondents shared included: Complexity of workload distribution across sites: 42% Latency between workloads and data at WAN edge: 39% Complexity of traffic prioritization: 36% Network congestion: 33% “It’s really not cheap to make your network AI ready,” McGillicuddy stated. “You might need to invest in a lot of new switches and you might need to upgrade your WAN or switch vendors. You might need to make some changes to your underlay around what kind of connectivity your AI traffic is going over.” Enterprise leaders intend to invest in infrastructure

CoreWeave acquires Core Scientific for $9B to power AI infrastructure push

Such a shift, analysts say, could offer short-term benefits for enterprises, particularly in cost and access, but also introduces new operational risks. “This acquisition may potentially lower enterprise pricing through lease cost elimination and annual savings, while improving GPU access via expanded power capacity, enabling faster deployment of Nvidia chipsets and systems,” said Charlie Dai, VP and principal analyst at Forrester. “However, service reliability risks persist during this crypto-to-AI retrofitting.” This also indicates that struggling vendors such as Core Scientific and similar have a way to cash out, according to Yugal Joshi, partner at Everest Group. “However, it does not materially impact the availability of Nvidia GPUs and similar for enterprises,” Joshi added. “Consolidation does impact the pricing power of vendors.” Concerns for enterprises Rising demand for AI-ready infrastructure can raise concerns among enterprises, particularly over access to power-rich data centers and future capacity constraints. “The biggest concern that CIOs should have with this acquisition is that mature data center infrastructure with dedicated power is an acquisition target,” said Hyoun Park, CEO and chief analyst at Amalgam Insights. “This may turn out to create challenges for CIOs currently collocating data workloads or seeking to keep more of their data loads on private data centers rather than in the cloud.”

CoreWeave achieves a first with Nvidia GB300 NVL72 deployment

The deployment, Kimball said, “brings Dell quality to the commodity space. Wins like this really validate what Dell has been doing in reshaping its portfolio to accommodate the needs of the market — both in the cloud and the enterprise.” Although concerns were voiced last year that Nvidia’s next-generation Blackwell data center processors had significant overheating problems when they were installed in high-capacity server racks, he said that a repeat performance is unlikely. Nvidia, said Kimball “has been very disciplined in its approach with its GPUs and not shipping silicon until it is ready. And Dell almost doubles down on this maniacal quality focus. I don’t mean to sound like I have blind faith, but I’ve watched both companies over the last several years be intentional in delivering product in volume. Especially as the competitive market starts to shape up more strongly, I expect there is an extremely high degree of confidence in quality.” CoreWeave ‘has one purpose’ He said, “like Lambda Labs, Crusoe and others, [CoreWeave] seemingly has one purpose (for now): deliver GPU capacity to the market. While I expect these cloud providers will expand in services, I think for now the type of customer employing services is on the early adopter side of AI. From an enterprise perspective, I have to think that organizations well into their AI journey are the consumers of CoreWeave.” “CoreWeave is also being utilized by a lot of the model providers and tech vendors playing in the AI space,” Kimball pointed out. “For instance, it’s public knowledge that Microsoft, OpenAI, Meta, IBM and others use CoreWeave GPUs for model training and more. It makes sense. These are the customers that truly benefit from the performance lift that we see from generation to generation.”

Oracle to power OpenAI’s AGI ambitions with 4.5GW expansion

“For CIOs, this shift means more competition for AI infrastructure. Over the next 12–24 months, securing capacity for AI workloads will likely get harder, not easier. Though cost is coming down but demand is increasing as well, due to which CIOs must plan earlier and build stronger partnerships to ensure availability,” said Pareekh Jain, CEO at EIIRTrend & Pareekh Consulting. He added that CIOs should expect longer wait times for AI infrastructure. To mitigate this, they should lock in capacity through reserved instances, diversify across regions and cloud providers, and work with vendors to align on long-term demand forecasts. “Enterprises stand to benefit from more efficient and cost-effective AI infrastructure tailored to specialized AI workloads, significantly lower their overall future AI-related investments and expenses. Consequently, CIOs face a critical task: to analyze and predict the diverse AI workloads that will prevail across their organizations, business units, functions, and employee personas in the future. This foresight will be crucial in prioritizing and optimizing AI workloads for either in-house deployment or outsourced infrastructure, ensuring strategic and efficient resource allocation,” said Neil Shah, vice president at Counterpoint Research. Strategic pivot toward AI data centers The OpenAI-Oracle deal comes in stark contrast to developments earlier this year. In April, AWS was reported to be scaling back its plans for leasing new colocation capacity — a move that AWS Vice President for global data centers Kevin Miller described as routine capacity management, not a shift in long-term expansion plans. Still, these announcements raised questions around whether the hyperscale data center boom was beginning to plateau. “This isn’t a slowdown, it’s a strategic pivot. The era of building generic data center capacity is over. The new global imperative is a race for specialized, high-density, AI-ready compute. Hyperscalers are not slowing down; they are reallocating their capital to

Arista Buys VeloCloud to reboot SD-WANs amid AI infrastructure shift

What this doesn’t answer is how Arista Networks plans to add newer, security-oriented Secure Access Service Edge (SASE) capabilities to VeloCloud’s older SD-WAN technology. Post-acquisition, it still has only some of the building blocks necessary to achieve this. Mapping AI However, in 2025 there is always more going on with networking acquisitions than simply adding another brick to the wall, and in this case it’s the way AI is changing data flows across networks. “In the new AI era, the concepts of what comprises a user and a site in a WAN have changed fundamentally. The introduction of agentic AI even changes what might be considered a user,” wrote Arista Networks CEO, Jayshree Ullal, in a blog highlighting AI’s effect on WAN architectures. “In addition to people accessing data on demand, new AI agents will be deployed to access data independently, adapting over time to solve problems and enhance user productivity,” she said. Specifically, WANs needed modernization to cope with the effect AI traffic flows are having on data center traffic. Sanjay Uppal, now VP and general manager of the new VeloCloud Division at Arista Networks, elaborated. “The next step in SD-WAN is to identify, secure and optimize agentic AI traffic across that distributed enterprise, this time from all end points across to branches, campus sites, and the different data center locations, both public and private,” he wrote. “The best way to grab this opportunity was in partnership with a networking systems leader, as customers were increasingly looking for a comprehensive solution from LAN/Campus across the WAN to the data center.”

Microsoft will invest $80B in AI data centers in fiscal 2025

And Microsoft isn’t the only one that is ramping up its investments into AI-enabled data centers. Rival cloud service providers are all investing in either upgrading or opening new data centers to capture a larger chunk of business from developers and users of large language models (LLMs). In a report published in October 2024, Bloomberg Intelligence estimated that demand for generative AI would push Microsoft, AWS, Google, Oracle, Meta, and Apple would between them devote $200 billion to capex in 2025, up from $110 billion in 2023. Microsoft is one of the biggest spenders, followed closely by Google and AWS, Bloomberg Intelligence said. Its estimate of Microsoft’s capital spending on AI, at $62.4 billion for calendar 2025, is lower than Smith’s claim that the company will invest $80 billion in the fiscal year to June 30, 2025. Both figures, though, are way higher than Microsoft’s 2020 capital expenditure of “just” $17.6 billion. The majority of the increased spending is tied to cloud services and the expansion of AI infrastructure needed to provide compute capacity for OpenAI workloads. Separately, last October Amazon CEO Andy Jassy said his company planned total capex spend of $75 billion in 2024 and even more in 2025, with much of it going to AWS, its cloud computing division.

John Deere unveils more autonomous farm machines to address skill labor shortage

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Self-driving tractors might be the path to self-driving cars. John Deere has revealed a new line of autonomous machines and tech across agriculture, construction and commercial landscaping. The Moline, Illinois-based John Deere has been in business for 187 years, yet it’s been a regular as a non-tech company showing off technology at the big tech trade show in Las Vegas and is back at CES 2025 with more autonomous tractors and other vehicles. This is not something we usually cover, but John Deere has a lot of data that is interesting in the big picture of tech. The message from the company is that there aren’t enough skilled farm laborers to do the work that its customers need. It’s been a challenge for most of the last two decades, said Jahmy Hindman, CTO at John Deere, in a briefing. Much of the tech will come this fall and after that. He noted that the average farmer in the U.S. is over 58 and works 12 to 18 hours a day to grow food for us. And he said the American Farm Bureau Federation estimates there are roughly 2.4 million farm jobs that need to be filled annually; and the agricultural work force continues to shrink. (This is my hint to the anti-immigration crowd). John Deere’s autonomous 9RX Tractor. Farmers can oversee it using an app. While each of these industries experiences their own set of challenges, a commonality across all is skilled labor availability. In construction, about 80% percent of contractors struggle to find skilled labor. And in commercial landscaping, 86% of landscaping business owners can’t find labor to fill open positions, he said. “They have to figure out how to do

2025 playbook for enterprise AI success, from agents to evals

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More 2025 is poised to be a pivotal year for enterprise AI. The past year has seen rapid innovation, and this year will see the same. This has made it more critical than ever to revisit your AI strategy to stay competitive and create value for your customers. From scaling AI agents to optimizing costs, here are the five critical areas enterprises should prioritize for their AI strategy this year. 1. Agents: the next generation of automation AI agents are no longer theoretical. In 2025, they’re indispensable tools for enterprises looking to streamline operations and enhance customer interactions. Unlike traditional software, agents powered by large language models (LLMs) can make nuanced decisions, navigate complex multi-step tasks, and integrate seamlessly with tools and APIs. At the start of 2024, agents were not ready for prime time, making frustrating mistakes like hallucinating URLs. They started getting better as frontier large language models themselves improved. “Let me put it this way,” said Sam Witteveen, cofounder of Red Dragon, a company that develops agents for companies, and that recently reviewed the 48 agents it built last year. “Interestingly, the ones that we built at the start of the year, a lot of those worked way better at the end of the year just because the models got better.” Witteveen shared this in the video podcast we filmed to discuss these five big trends in detail. Models are getting better and hallucinating less, and they’re also being trained to do agentic tasks. Another feature that the model providers are researching is a way to use the LLM as a judge, and as models get cheaper (something we’ll cover below), companies can use three or more models to

OpenAI’s red teaming innovations define new essentials for security leaders in the AI era

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More OpenAI has taken a more aggressive approach to red teaming than its AI competitors, demonstrating its security teams’ advanced capabilities in two areas: multi-step reinforcement and external red teaming. OpenAI recently released two papers that set a new competitive standard for improving the quality, reliability and safety of AI models in these two techniques and more. The first paper, “OpenAI’s Approach to External Red Teaming for AI Models and Systems,” reports that specialized teams outside the company have proven effective in uncovering vulnerabilities that might otherwise have made it into a released model because in-house testing techniques may have missed them. In the second paper, “Diverse and Effective Red Teaming with Auto-Generated Rewards and Multi-Step Reinforcement Learning,” OpenAI introduces an automated framework that relies on iterative reinforcement learning to generate a broad spectrum of novel, wide-ranging attacks. Going all-in on red teaming pays practical, competitive dividends It’s encouraging to see competitive intensity in red teaming growing among AI companies. When Anthropic released its AI red team guidelines in June of last year, it joined AI providers including Google, Microsoft, Nvidia, OpenAI, and even the U.S.’s National Institute of Standards and Technology (NIST), which all had released red teaming frameworks. Investing heavily in red teaming yields tangible benefits for security leaders in any organization. OpenAI’s paper on external red teaming provides a detailed analysis of how the company strives to create specialized external teams that include cybersecurity and subject matter experts. The goal is to see if knowledgeable external teams can defeat models’ security perimeters and find gaps in their security, biases and controls that prompt-based testing couldn’t find. What makes OpenAI’s recent papers noteworthy is how well they define using human-in-the-middle

Stay Ahead, Stay ONMINE