Год: 2025

When Lightbulb AI was founded in Mumbai in 2021 by Vishal Soni, Ritu Soni Srivastava, and Yogesh Sachdeva, the team wasn’t setting out to reinvent the global market research industry. In fact, their earliest experiments had little to do with advertising or consumer insights at all.

“Initially, Lightbulb came about as an idea in attention capture for the EdTech vertical,” Vishal Soni, chief product officer at Lightbulb AI, told AIM. The hypothesis was simple: as education moved online during the pandemic, could technology measure and improve student attention?

But as post-pandemic behaviour shifted and EdTech adoption slowed, the founders discovered an unexpected pull from the world of market research, where brands were struggling to understand not just what people said, but how they felt. Lightbulb’s emotion-tracking and eye-tracking models, built originally for children’s learning, suddenly unlocked groundbreaking new commercial applications.

“What happened after COVID was that the hypothesis on EdTech… didn’t pan out as we expected. But people wanted to understand how consumers were emoting when exposed to a particular stimulus,” Soni explained. This pivot proved decisive.

Today, Lightbulb AI stands at the forefront of insight technology, offering an integrated platform that combines eye-tracking, facial coding, AI-driven report generation, and generative AI-powered analysis.

And brands are taking notice, including Myntra, Merrill Research (US), and TVS Motors, with whom Lightbulb has collaborated. “We have submitted papers with these customer partners at the forums. So it’s on public record,” the founders confirmed.

Soni and Sachdeva’s partnership predates Lightbulb by nearly two decades. “We were together in our last venture also… and then we started this new venture,” said Sachdeva, who is the CTO.

This long-term collaboration helped them evolve rapidly into a deep tech research company. Their previous exit to RoundGlass also equipped them with operational maturity uncommon for early-stage founders.

Eye-Tracking, Facial Coding

Lightbulb’s platform blends quantitative and qualitative research capabilities in ways that traditional agencies rarely achieve. For quantitative work, the system analyses participants’ gaze patterns through eye-tracking, maps their emotional responses using facial coding, and incorporates survey data enhanced through generative AI. As Soni explained, “On the quantitative research aspect, we do eye-tracking… emotion mapping… and online surveys. All these things together give you the quant solution.”

On the qualitative side, innovation is even more fundamental. Traditional focus groups can take weeks to analyse, but Lightbulb compresses this entire process into under an hour. “A report writing which would take a week can now happen in 30 minutes… and presentations take another 10 minutes. Within 45 minutes to an hour, you’re done,” Soni noted.

Behind the scenes, AI continually improves accuracy. Sachdeva admitted that “Improving accuracy for emotions… other emotions have less raw data, and that is a major challenge,” but the team has trained its proprietary models on an enormous dataset of 9 million faces, making it one of India’s most robust emotion-recognition systems.

Given that Lightbulb processes video, facial expressions, and gaze data, privacy is paramount. Sachdeva emphasised that the company is fully compliant with international standards, including ISO 27001, ISO 27701, and GDPR. “No biometric information is stored, faces are mapped only to participant IDs, and data is deleted after six months to a year,” he explains.

Soni added that on the user-facing side, “We take explicit consent from the user… permission for camera, eye-tracking, and analysis at every stage.” This strict adherence to global standards has made Lightbulb a trusted partner for enterprise clients.

Global Recognition, and the Clients Behind It

Lightbulb AI has raised $1.5 million in seed funding from Chiratae Ventures. According to Soni, the team is not pursuing additional rounds immediately, choosing instead to strengthen product-market fit, stabilise recurring revenue, and move toward breakeven before exploring Series A opportunities. “We felt that we are capitalised enough to power through our Indian strategy,” he said, adding that the firm would look at fundraisers after another year, given its deliberate approach to sustainable growth.

Lightbulb AI’s rapid rise is reflected in industry accolades. The company became a finalist at MRSI and ESOMAR, won Gold at ESOMAR APAC for its work with Myntra, and was selected among the Top 50 deep tech startups at Singapore’s Slingshot Challenge. It was also one of only two Indian startups shortlisted at Hong Kong’s HKSTP startup competition.

These wins were driven by real deployments with clients such as Myntra, Merrill Research, and TVS Motors, spanning industries from retail to automotive to global research consulting. Each has used Lightbulb to decode human attention and emotion through advanced AI, demonstrating measurable improvements in insight depth and speed.

Market Momentum and the AI Productivity Mandate

According to Soni, “2025 and ’26 are watershed moments for productivity enhancement.” Companies across sectors now face pressure to do more with leaner teams, and Lightbulb’s value proposition—faster research, deeper insight, and lower operational cost—aligns directly with that mandate. He added that the industry-wide question today is, “Are you doing something with AI to enhance productivity?”

This demand is reflected in Lightbulb’s retention numbers. “We’ve had very little churn… some clients are in their third or fourth renewal cycle,” Soni noted, emphasising the stickiness of the platform.

Lightbulb began with a project-based model but has increasingly shifted toward a subscription paradigm. “The subscription model has worked really well for us… this year we’ve seen a lot of uptake,” said Soni.

Their growth strategy is unfolding in phases. After consolidating their presence in India, the team plans to expand into the US by deploying dedicated marketing and outreach personnel.

Simultaneously, they are building joint partnerships in Southeast Asia, tapping into established research networks. “We want somebody established there to take us as a technology provider along with them,” Soni explained.

The Challenges Ahead, and Why Lightbulb Isn’t Afraid

The founders are realistic about the competitive landscape. Sachdeva observed that as mainstream platforms like Google and OpenAI improve, and let people generate reports directly, they may not be interested in companies like theirs. Yet, he sees this as an opportunity, too: “Once they compare, they see the value we create beyond generic LLM outputs.”

Soni pointed to a larger existential challenge. “The speed at which OpenAI and Gemini are executing… that’s a major challenge for any organisation that relies on AI models,” he said, noting that large foundational models can generate code, images, videos, and analysis at a pace that reshapes entire industries.

Despite this, Lightbulb retains an advantage: deep specialisation in multimodal research workflows, something generic AI tools cannot replicate without domain-specific engineering, contextual tuning, and integrated gaze and emotion capabilities.

A Future Built on Multimodal Intelligence

Lightbulb’s next breakthrough aims to bring together three historically siloed data streams: eye-tracking, facial coding, and survey data. “We’re working on integrating these to bring out highly meaningful unified reports… that’s the challenge we’re really excited about,” Soni said.

This vision reflects a broader shift toward AI-native insight generation, where multimodal signals converge into an automatically generated narrative about consumer behaviour, richer, faster, and more accurate than anything created manually.

In four short years, Lightbulb AI has evolved from a pandemic-era EdTech experiment into one of India’s most advanced insight platforms, trusted by global brands, awarded by international research bodies, and recognised among deep-tech innovators.

Its founders are aware of the challenges ahead, but they remain grounded in the belief that understanding human attention and emotion at scale is a universal need, one that AI, when implemented responsibly and creatively, is uniquely positioned to solve.

If the last few years are any indication, Lightbulb AI isn’t just illuminating consumer behaviour. It’s lighting the path for the future of the research industry itself.

The post This Mumbai-based Startup is Illuminating the Future of Consumer Understanding appeared first on Analytics India Magazine.

IT services company Mphasis’ Sparkle Innovation Program has emerged as a significant channel for enterprise-focused innovation, helping it accelerate solution development, strengthen its AI-led portfolio, and support clients across banking, insurance, healthcare, logistics, and other sectors.

Launched in 2016, the initiative aims to foster innovation by partnering with startups, academic bodies and research entities. Over the years, Mphasis has partnered with ventures across sectors, including fintech Upswot to cater to US-based regional and global banks, as well as the Museum of Art & Photography through its CSR arm.

In an email interaction with AIM, Srikumar Ramanathan, chief solutions officer at Mphasis, said the programme is delivering measurable business outcomes and is set to expand further over the next two years.

Ramanathan said Sparkle has shortened evaluation cycles, reduced operational costs and increased pipeline velocity by enabling differentiated co-innovated offerings. He noted that several Sparkle-led engagements have matured into multi-year contracts across industries, including banking, insurance, healthcare, and supply chain.

The programme, he said, has helped clients adopt AI-driven, cloud-native and digital transformation initiatives more rapidly, creating new revenue streams, improving efficiency and enhancing competitiveness.

Through the program, Mphasis has been actively collaborating with Nasscom’s InnoTrek participants, particularly US-based startups, by offering mentorship, enterprise solution guidance, and customer access.
InnoTrek aims to help selected startups achieve global visibility and scale by supplying the necessary knowledge, tools, and connections.

In the 2025 edition of the programme, Mphasis recently partnered with five standout startups from the InnoTrek cohort—Edgeable AI, Perpetuuiti Technosoft, QuoQo, and SuperBryn AI—helping them refine their offerings and connect with enterprise clients worldwide.

The company said it will be working closely with these startups to co-create enterprise-grade solutions by providing technical mentorship, refining their go-to-market strategies, and facilitating access to global enterprise clients.

PoCs to Full-Scale Production

The CSO said a significant number of proofs-of-concept have moved into full-scale production deployments.

These span digital banking back-office automation, data intelligence, fraud detection, workflow digitisation, underwriting and claims-processing automation in insurance, patient data analytics and digital workflows in healthcare, as well as process automation and supply-chain visibility solutions in logistics and manufacturing.

He added that Sparkle also supports cybersecurity and IT resilience use cases, including threat detection, risk monitoring, and compliance automation.

Across sectors, Mphasis’ co-innovated solutions have delivered quantifiable results, Ramanathan asserted.

In insurance, cognitive and automation platforms improved underwriting and claims turnaround while lowering manual processing costs. In banking, AI-enabled automation reduced dependency on legacy systems and improved onboarding and service efficiency.

Healthcare clients, he said, saw better patient data management and more efficient care coordination. Logistics and supply-chain firms benefited from modernisation and productivity improvements, while cybersecurity clients improved their security posture and remediation cycles.

“Across these examples, clients saw faster deployment, improved efficiency, reduced manual effort, and better scalability,” he noted.

In Q2 FY26, the company reported new deal wins (TCV) worth $528 million, with 87% coming from new-gen services such as cloud, AI, and digital engineering. During the quarter, it scored six large deals, including one contract over $100 million with a BFSI client and two above $50 million, underscoring sustained enterprise-level demand.

This followed a strong Q1 FY26, where Mphasis logged $760 million in TCV wins, its highest quarterly figure in recent times.

Ramanathan said Sparkle acts as a feeder into the company’s AI-led, platform-driven portfolio.

By validating partner technologies in sandboxed environments and integrating them into Mphasis’ Front2Back architecture and X2C² framework for quick cloud adoption and cognitive transformation, the company has expanded its AI services more rapidly.

This includes document intelligence, voice AI, cognitive automation, and predictive analytics.

He said time from concept to deployment has “shrunk significantly,” enabling capabilities such as digital underwriting, cloud-native banking workflows, healthcare data platforms and supply-chain automation to reach production faster.

Selection Process

Mphasis selects partners based on alignment with client pain points, technical maturity, enterprise readiness, and integration capability. The CSO said scalability, cloud-native design, compliance standards and complementarity with Mphasis’ frameworks are essential criteria.

Once selected, technologies undergo technical due diligence, sandbox prototyping, use-case mapping and hardening before being integrated into Mphasis’ reference architectures and deployed with enterprise-grade support and SLAs.

Scaling innovations globally comes with challenges such as regulatory differences, legacy system complexity, performance scale-up, and security and compliance requirements. Mphasis addresses these through early compliance assessments, standardisation via its architectures, stress-testing, and leveraging the company’s global delivery footprint and domain-specific Centres of Excellence (CoE), Ramanathan said.

Sparkle differentiates itself by being “client problem–first” and embedding co-innovated solutions directly into enterprise delivery pipelines through Front2Back and X2C², he added.

He said the model is outcome-driven and enables future-ready, cloud-native, AI-led and compliance-focused modernisation for clients.

Ramanathan claimed the company is seeing strong enterprise traction in AI-led automation, cloud-native modernisation, cybersecurity and compliance automation, voice AI, and supply-chain and logistics automation. These areas reflect rising demand for resilience, scalability and better customer experience, particularly in regulated sectors.

Evaluating the programme, he said Sparkle has matured into a “dependable innovation channel” that bridges emerging technologies with real business needs.

Over the next 12–24 months, Mphasis expects to scale more PoCs to global production, deepen investments in AI-native and cloud-native platforms, expand domain-specific blueprints, strengthen global delivery CoEs, and support growing demand for integrated AI, automation, cloud and cybersecurity solutions.

“Sparkle is evolving into a robust, enterprise-ready, multi-industry innovation ecosystem that supports clients’ digital transformation journeys end-to-end,” he signed off.

The post How Sparkle Program Is Winning Multi-Year Deals for Mphasis appeared first on Analytics India Magazine.

Valued at $750 billion, OpenAI today rivals technology companies that took decades to build. What began as a nonprofit research lab has evolved into a full-stack AI company, spanning frontier models, consumer and enterprise applications, compute infrastructure and AI hardware.

By comparison, Microsoft, founded in 1975, now trades at a valuation above $3 trillion after more than 40 years in business. Alphabet, Google’s parent company, became one of the fastest firms to surpass a multi-trillion-dollar value roughly two decades after its 2004 IPO.

Amazon, launched in 1994, followed a similar arc, building its valuation over years of expansion in cloud computing and e-commerce.

OpenAI’s rise has unfolded at a dramatically faster pace. As the company completes 10 years in December, CEO Sam Altman reflected on its trajectory in a blog post.

“OpenAI has achieved more than I dared to dream possible; we set out to do something crazy, unlikely, and unprecedented,” he wrote.

Altman is explicit about what comes next. He said he believes OpenAI is “almost certain” to build superintelligence within the next decade. The transformation, he suggested, may feel subtle on the surface. “The people of 2035 will be capable of doing things that I just don’t think we can easily imagine right now.”

Tyler Johnston, executive director of the nonprofit Midas Project, described OpenAI’s growth as historically compressed. “OpenAI is a company that has existed on this insanely accelerated lifespan,” Johnston told TechBrew, adding that the scale of change within the organisation over a decade has been “astonishing”.

“It’s just astonishing how much has happened in the 10 years that it’s existed, and how many different visions of the organisation have, at various times, flourished or been squashed based on how convenient they were to the leadership at the time.”

From boardroom tussles and leadership reshuffles to escalating rivalry with Google and other tech giants, OpenAI’s journey has resembled a high-stakes battle. More recently, the company has declared a “code red” competitive posture, culminating in the release of its latest flagship model, GPT-5.2.

So how did it all begin?

In December 2015, a group of researchers and investors, including Altman, Greg Brockman, Ilya Sutskever, Elon Musk, Peter Thiel and Reid Hoffman, pledged $1 billion to launch a nonprofit lab.

There was no product roadmap, no revenue model and no certainty of survival. What guided it instead was the conviction that AI would reshape the world, and that its direction could not be left unchecked.

A decade later, OpenAI has become a central pillar of the global AI ecosystem, being one of the most influential companies in the world. Its models now power startups, enterprises, governments, classrooms, creative studios and developer ecosystems.

As Daleep Singh, former deputy national security adviser and head of global macroeconomic research at PGIM, puts it, OpenAI’s influence has grown so pervasive that “if OpenAI falters, the foundations for the entire AI sector become fragile.” He told Axios, “You have to think about the financial contagion.”

The Early Years: Research Before Revenue

From 2016 to 2018, OpenAI looked like a traditional research lab. It released open tools like Gym and Universe, trained reinforcement-learning agents to play Atari games, and published robotics breakthroughs such as a robotic hand capable of solving a Rubik’s Cube.

In 2018, Musk stepped down from OpenAI’s board, citing potential conflicts with Tesla’s AI efforts. It later emerged that he had proposed taking control of OpenAI and merging it with Tesla, believing the company was falling behind Google.

Despite the turbulence, the research did not slow down. The first mainstream breakthrough came in 2019 with OpenAI Five, which defeated professional human players at Dota 2. The feat proved that large-scale learning systems could master complexity far beyond scripted rules.

That same year, OpenAI released GPT-2, but delayed its full release citing safety concerns. The move sparked controversy and set the tone for how the company would balance progress with risk.

In 2020 came GPT-3, a 175-billion-parameter language model that stunned developers. It could write essays, generate code, answer questions and mimic styles with uncanny fluency.

ChatGPT: When AI Went Mainstream

Then came November 2022.

ChatGPT was released as a research preview, powered by GPT-3.5. Within days, it spread faster than any consumer app in history. For the first time, AI felt conversational, accessible and personal. OpenAI soon followed with subscription plans, launching ChatGPT Plus and ChatGPT Enterprise.

By 2023, OpenAI had crossed into the multi-billion-dollar annual revenue, driven by subscriptions, API usage and enterprise licensing.

The Boardroom Crisis That Nearly Broke OpenAI

In late 2023, OpenAI’s board abruptly removed Sam Altman as CEO, citing governance concerns. Within days, employees revolted, Microsoft intervened, and Altman was reinstated. The episode exposed deep tensions inside the company—between safety and speed, research and commercialisation, nonprofit oversight and market pressure.

In the aftermath, OpenAI restructured its board, consolidated leadership and began moving towards a more conventional corporate setup. By 2025, OpenAI formally transitioned into a public-benefit corporation, with a new nonprofit foundation holding a significant ownership stake.

Beyond Text and GPT-4o

OpenAI never intended to remain text-only. In January 2021, the launch of DALL·E showed that models could generate images from imagination. Whisper soon followed, tackling speech recognition.

Then, in 2024, it unveiled Sora, a text-to-video model capable of producing cinematic clips from simple prompts. However, the defining release of the year was GPT-4o, OpenAI’s first natively multimodal model. It brought real-time voice interaction that allowed users to speak with the model in natural, low-latency conversations.

In 2025, OpenAI introduced native image generation in GPT-4o, a release that quickly went viral due to the Ghibli-style image trend across social platforms.

And that wasn’t all.

This year also saw the launch of Sora 2 alongside a standalone app, as well as a partnership with Disney, opening access to its vast catalogue of characters and worlds to generate.

GPT-5, GPT-5.1, GPT-5.2…

GPT-5 arrived at a moment when OpenAI’s strategy was already clear. Released in 2025, it was presented as a consolidation model. Instead of fragmenting capabilities across dozens of specialised systems, OpenAI positioned GPT-5 as a single system that could write, reason, use tools, analyse images and carry context across long sessions.

GPT-5.2 is the company’s latest launch, which it claims is its best model for professional work.
Over the past couple of years, OpenAI has also introduced tools beyond ChatGPT, including Codex for AI-assisted programming, Atlas browser for exploring large and structured knowledge sources, and reasoning models such as the o1 series, designed to handle complex, multi-step problem-solving rather than simple text prediction.

Business Growth, Partnerships and Valuation

OpenAI’s business model and partnerships evolved rapidly. What began in 2020 as an experiment with APIs has grown into a business powered by subscriptions, enterprise deals and developers building on top of its models.

According to Altman, OpenAI is on track to cross a $20 billion annualised revenue run rate by the end of this year. That growth, however, carries a cost that few companies in history have had to confront.

Modern AI does not scale quietly.

To meet rising demand, OpenAI is planning infrastructure investments of roughly $1.4 trillion over the next eight years. That ambition is now taking physical shape through Stargate, the company’s long-term effort to build next-generation AI data centres for frontier models.

OpenAI has raised about $64 billion overall, with the March $40 billion round led by SoftBank, Microsoft, Thrive Capital, Dragoneer, Coatue and Altimeter.

The Messy Middle of the OpenAI Story

As OpenAI’s reach expanded, so did the list of people willing to challenge it.

The most serious pressure came through the courts. In late 2023 and 2024, authors including George RR Martin, John Grisham and Sarah Silverman, along with major publishers, filed lawsuits accusing OpenAI of training its models on copyrighted books without permission. The New York Times followed with its own lawsuit, alleging that OpenAI models could reproduce large portions of its articles and were built using its journalism without authorisation.

The legal risk was significant. By 2024, the company began signing licensing deals with media groups such as News Corp, choosing compromise over conflict.

The same year brought internal change. Several senior leaders and researchers left the company, including co-founder and chief scientist Sutskever and chief technology officer Mira Murati.

What’s Next?

A decade in, OpenAI sits at an unusual crossroads. It is no longer just a research lab wrestling with abstract questions about AI, nor is it a conventional technology company chasing incremental growth. It has become a platform, an infrastructure provider and, increasingly, a reference point for how AI power is built, governed and distributed.

The coming years will test whether OpenAI can hold together its founding ideals and its present-day realities. The push toward superintelligence, the scale of capital and infrastructure required, and the growing scrutiny from regulators, courts and competitors all raise questions that cannot be answered by model releases alone.

The post How OpenAI Became the Most Valued AI Company in 10 Years appeared first on Analytics India Magazine.

NVIDIA on Monday announced the NVIDIA Nemotron 3 family of open models, datasets and libraries aimed at supporting the development of transparent and efficient multi-agent AI systems across industries.

The Nemotron 3 lineup includes Nano, Super and Ultra models built on a hybrid latent mixture-of-experts (MoE) architecture, which NVIDIA says is designed to reduce inference costs, limit context drift and improve coordination among multiple AI agents.

“Open innovation is the foundation of AI progress,” NVIDIA founder and CEO Jensen Huang said. “With Nemotron, we’re transforming advanced AI into an open platform that gives developers the transparency and efficiency they need to build agentic systems at scale.”

Among the three models, Nemotron 3 Nano is available immediately. It is a 30-billion-parameter model that activates up to 3 billion parameters per task and is optimised for low-cost inference use cases such as software debugging, summarisation and AI assistants. NVIDIA said the model delivers up to four times higher token throughput than Nemotron 2 Nano and reduces reasoning token generation by up to 60%.

It is available on Hugging Face and through inference providers such as Baseten, DeepInfra, Fireworks, FriendliAI, OpenRouter and Together AI. The model is also offered as an NVIDIA NIM microservice for deployment on NVIDIA-accelerated infrastructure.

Nemotron 3 Nano will also be available on AWS via Amazon Bedrock and supported on multiple cloud platforms in the coming months.

On the other hand, Nemotron 3 Super is a roughly 100-billion-parameter model, designed for multi-agent applications requiring low latency, while Nemotron 3 Ultra, with about 500 billion parameters, is intended for deep reasoning and long-horizon planning tasks.

Both Super and Ultra use NVIDIA’s 4-bit NVFP4 training format on Blackwell GPUs to reduce memory requirements. These models are expected to be available in the first half of 2026.

The launch comes as companies move beyond single AI chatbots toward collaborative agent-based systems, where multiple models work together on complex workflows.

According to NVIDIA, Nemotron 3 allows developers to route tasks between frontier proprietary models and open Nemotron models within the same workflow to balance reasoning capability and cost efficiency.

NVIDIA said the Nemotron 3 family also aligns with its sovereign AI strategy, allowing governments and enterprises to deploy models tailored to local data, regulations and policy requirements. Organisations across Europe and South Korea are among those adopting the open models, the company said.

Several enterprise customers and partners, including Accenture, Deloitte, EY, Oracle Cloud Infrastructure, Palantir, Perplexity, ServiceNow, Siemens, Synopsys and Zoom, are integrating Nemotron models into AI workflows spanning manufacturing, cybersecurity, software development and communications.

Perplexity CEO Aravind Srinivas said the company is using Nemotron within its agent routing system to optimise performance. “We can direct workloads to fine-tuned open models like Nemotron 3 Ultra or use proprietary models when tasks require it,” he said.

Alongside the models, NVIDIA released three trillion tokens of pretraining, post-training and reinforcement learning datasets, including an Agentic Safety Dataset for evaluating multi-agent systems. The company also open-sourced NeMo Gym, NeMo RL and NeMo Evaluator to support training, customisation and evaluation of agentic AI.

The post NVIDIA Unveils Nemotron 3 Open Models to Power Multi-Agent AI Systems appeared first on Analytics India Magazine.

Physical AI is often described as the next great leap for AI, with many investors and companies expecting a sudden breakthrough, similar to what ChatGPT did for language models.

However, the wait for an I, Robot-like future, where advanced AI integrates with robots, vehicles, and other devices, may be long. During a panel discussion at AWS re:Invent 2025 conference in Las Vegas, industry stakeholders agreed that Physical AI will not arrive as a single breakthrough, viral demo, or dominant model. Physical AI, they argued, is constrained by realities that digital AI never faced.

On the panel, Ryu Jung-hee, founder and chief executive officer of South Korean robotics startup RLWRLD, was joined by Kevin Peterson, CTO of Bedrock Robotics; Sri Elaprolu, director of AWS Generative AI Innovation Centre; Amit Goel, head of robotics and edge computing ecosystem at NVIDIA; and Josh Gruenstein, co-founder and CEO of Tutor Intelligence.

At the centre of this discussion was the idea that intelligence alone is no longer the hard problem. Once AI systems are expected to move, grip, and act in real environments, speed, safety, data, and hardware begin to dictate progress—and that’s where consumer expectations lie.

In an exclusive conversation with AIM on the sidelines of the event, Jung pitched that physical AI will not arrive through one viral model or a single company. Instead, it will emerge unevenly, shaped by hardware limits, real-time control, and regional data advantages.

Unlike ChatGPT, which improved rapidly by scaling data and compute, Physical AI must solve multiple problems at once. “There are two challenges,” Jung said. “Number one is data. Number two is the embodiment, the robotics system, the hardware.”

“Action is not so simple,” Jung said. “They should control the robot itself, and in real-time, because existing LLMs or vision language models are not real-time models.”

At a time when humanoid demos are entering the market, with 1X’s NEO bipedal humanoid robot capable of chores like laundry and cleaning, Jung offers a grounded counterpoint. He believes intelligence is no longer the main constraint; the harder problem is turning perception and reasoning into safe, fast, instantaneous action.

Thinking is Easy, Acting is Not

Several speakers at the panel asserted that Physical AI behaves differently from digital systems.

Goel said digital AI spread overnight because it could reach millions of users instantly. “Physically, AI is different,” he said. “There are many open challenges… that need to be solved.”

One challenge is the sheer volume of data the physical world generates. “We have to understand force. We have to understand audio,” Goel said. “The amount of data generated from the physical world obviously differs in magnitude—more than what the text data exists.” That data cannot simply be scraped from the internet. It must be produced through robots operating in factories, warehouses, and construction sites.

Jung added that robots require real-world movements that demand immediate decisions. “It also has to be fast in real-time,” he said, explaining why existing AI models cannot directly control physical systems.

ChatGPT scaled because text already existed, and mistakes carried comparatively little risk. Physical AI must deal with motion, balance, and force. Errors break machines and harm people. “How can we add in some real-time model, a real-time action model to the existing vision language model?” Jung pondered.

Even when models are trained, they cannot be deployed casually. “Unlike the digital world, where you can just wide check your models, how do you wide check a physical model? You need a simulator,” Goel enlightened. Verification and validation become central, not optional.

Dexterity is Essential

Jung noted that consumers don’t understand the complexity in integrating AI with hardware. “But we have a long way to go.”

The challenge becomes sharper when it comes to dexterity. Human hands use many joints at once. Most robots still rely on simple grippers. Jung noted that it is not a design choice but a limit of current systems.

“The hardest thing is controlling their high degree of freedom,” he said. “Providing some high level of dexterity means controlling the multiple joints of the humanoid.”

A high degree of freedom robot is one with seven or more robotic axes that offer an enhanced range of motion.

Many robots still rely on simple grippers. “None of them is now dealing with the high level of freedom hardware yet,” Jung said. “That’s why they stuck to the two fingers.” He added that even well-funded efforts struggle. “Even the Tesla Optimus team cannot provide better hardware, especially hands. They fail to [offer] high degree of freedom.”

RLWRLD is building robot hands with 15 degrees of freedom to handle complex tasks.

The reason, he argued, is not ambition but feasibility. Each extra joint multiplies the data and control space. Training those systems requires new datasets and new models. “Between the data and the hardware, we should develop the foundation model to control the robot,” Jung said.

Scaling by Region, Not Virality

Another reason Physical AI will not have a single defining moment is geography. Digital AI globalised quickly because it relied on shared infrastructure. Physical AI depends on supply chains, labour, and industrial data—all of which are unevenly distributed.

“The US is the king of software. China is the king of hardware,” Jung said in the interview. Between them sit countries like South Korea and Japan, which hold an advantage in industrial data. “Number one strength we have is industrial data,” he said, referring to East Asian manufacturing ecosystems.

India enters this picture for different reasons. “Humanoid cannot assemble humanoid,” Jung said. “Humanoid should be assembled by humans.” That labour intensity pushes companies to look for manufacturing alternatives outside mainland China. “The manufacturing inside of mainland China is not widely adopted by the US allies,” he said. “India is, I think, one of the best alternatives.”

Panellists echoed the idea that Physical AI will grow through ecosystems rather than platforms. Goel painted a hybrid future where intelligence is split between cloud and edge systems.

“You cannot be dependent on the cloud connectivity for new latency tasks,” he said. “You have to have compute on the edge.”

This architecture makes it less likely for a single model to become dominant. Instead, progress depends on how well companies assimilate training, simulation, deployment, and safety.

Gruenstein added that even industrial settings resist standardisation. “99% manufacturers in the United States are small businesses,” he said. “They can’t afford any of that stuff.” Many facilities change tasks daily, which breaks traditional automation. That variability forces robotics companies to adapt systems to local conditions rather than chase universal solutions.

This regional spread makes a single winner unlikely. Physical AI systems must adapt to local data, regulation, and infrastructure. Jung said even large US companies cannot operate alone. “In the US, even the big giants like AWS cannot survive alone,” he said. “They need some ecosystem from hardware to AI.”

Not One Single Moment

Public perception often overestimates the readiness of humanoid robots, given how digital AI products collect feedback from millions of users and iterate. Robotics cannot do the same. That’s where scalability also becomes an issue, Jung argued.

Peterson also agreed that progress would be gradual. “We’re going to see applications that roll out over time,” he said. “And then, in five years, we’re going to look back and say a lot of this work is easier.”

Some startups are pushing early robots into homes to gather data. Jung questioned that approach. “That is a very selfish idea because from the point of an end user, the product itself is not complete,” he said. He warned that such strategies trade user trust for speed.

Instead, he expects gradual progress, driven by industry use cases and regional ecosystems. “Sooner or later, we can provide a better model,” Jung predicted. “I’m very positive about providing better architecture in this market.”

For an industry searching for its ChatGPT moment, the message from founders and engineers is consistent: there may never be one.

The post Why Physical AI Will Never Have its ChatGPT Moment appeared first on Analytics India Magazine.

While Bengaluru leads in talent and technology, Mumbai recorded the highest office leasing by BFSI GCCs, according to Savillis India’s latest report, Global Capability Centres: Enabling India’s Strategic Advantage.

This reflects the strength of the financial services ecosystem in India’s commercial capital.

Savills noted that Mumbai benefits from a well-established network spanning primary, secondary and tertiary financial services, making it a natural choice for BFSI-focused global centres.

Interestingly, Mumbai witnessed the highest office leasing by BFSI GCCs, even though Bengaluru leads in talent in the segment.

Bengaluru has emerged as the undisputed leader in India’s GCC real estate landscape, accounting for 42.8 million sq. ft. of office leasing between 2020 and 2024, or 38% of the national total, according to the report.

The city anchors the country’s top three GCC markets, along with Hyderabad and Pune.

Together, the three regions drove nearly 70% of India’s total GCC absorption of 112 million sq. ft. during the period, growing at a 12% CAGR, the report said.

Savills attributes Bengaluru’s dominance to its deep technology ecosystem and talent pipeline, supported by over 2.2 million STEM graduates annually.

The city continues to lead in IT-BPM GCCs, commanding 77% of leasing among the top three cities. In engineering and manufacturing, Bengaluru, Pune and Chennai together account for 78% of demand.

Looking ahead, India’s GCC office demand is projected to reach 180 million sq. ft. between 2025 and 2030 under a realistic 8.2% CAGR.

Emerging sectors such as automotive, life sciences and semiconductors are expected to further boost demand, particularly in technology-led cities.

“Green infrastructure, hybrid work models and strategic location choices will characterise the next wave of GCC expansion,” said Arvind Nandan, MD, research and consulting, Savills India.

“In India, GCC real estate is no longer just treated as a cost, but as an investment to help businesses attract talent and innovate.”

Echoing this shift, Naveen Nandwani, MD, commercial advisory & transactions, Savills India, said, “ The shift toward high-value technology, digital and innovation functions is driving the demand for future-ready workspaces.”

India currently hosts around 1,800 GCCs employing 1.9 million professionals, a number expected to rise to 2,200 GCCs and 2.8 million employees by 2030.

Core sectors such as software & IT services, BFSI, engineering & manufacturing, pharma, retail and consumer services continue to dominate, while newer clusters are gaining traction.

GCC roles in India also command a premium, with salaries 12–20% higher than those in traditional IT services, driven by demand for skills in AI/ML, data engineering, cybersecurity, intelligent automation, and cloud platforms.

Savills expects leasing momentum to remain strong, with GCCs likely to absorb around 30 million sq. ft. annually between 2025 and 2030.

The post Mumbai Tops Office Leasing for BFSI GCCs Despite Bengaluru’s Talent Edge appeared first on Analytics India Magazine.

Cognizant has launched a Cyber Defence Centre (CDC) in Bengaluru, strengthening its global cybersecurity operations and expanding its ability to deliver advanced, AI-powered cybersecurity services to clients worldwide.

In a release, the company said the Bengaluru CDC is the company’s largest such facility, positioning it as a key hub within its global network of cyber defence centres.

It is designed to provide platform-centric cybersecurity managed services, along with cybersecurity engineering and transformation services, to enterprises across industries and geographies.

The company said the new centre is staffed by experienced cybersecurity professionals who will offer round-the-clock monitoring and rapid incident response.

To ensure a steady pipeline of skilled professionals, Cognizant said the centre has partnered with leading academic institutions to train next-generation cybersecurity talent.

The facility also houses an integrated threat research lab and is supported by leading technology companies.

“Cybersecurity is entering a new era where hyper-connected enterprises and AI-driven threats outpace traditional defences,” said Ravi Kumar S, CEO, Cognizant, stated in a release.

“Our new Cyber Defence Centre in Bengaluru reflects our commitment to delivering intelligence-led cybersecurity engineering and operation services that help enable clients to predict, prevent and mitigate advanced cybersecurity threats.”

According to Cognizant, the Bengaluru CDC is built on a multi-layered technology architecture that brings together the company’s proprietary innovations, including Cognizant Neuro Cybersecurity, with industry-leading platforms from technology partners to deliver advanced cybersecurity capabilities.

With the addition of the Bengaluru facility, the company said it is expanding the scale and depth of its global cyber defence centre network, aiming to deliver comprehensive enterprise cybersecurity services that help clients build risk resilience while advancing innovation in cyber defence strategies.

The post Cognizant Opens Cyber Defence Centre in Bengaluru appeared first on Analytics India Magazine.

As technology leaders, we all share the same mandate — “Do more with less.” In the past, it meant streamlining infrastructure or optimising headcount. But in today’s era of data and AI, productivity is not measured by your efficiency but by making every employee and every system work smarter.

Here are six practical lessons that any CTO can apply to drive measurable enterprise productivity.

1. Fix Your Data Quality and Governance First

No AI, automation, or analytics initiative can outperform poor data quality. Organisations lose 20–30% of revenue each year due to bad data such as duplicate records, missing fields, outdated customer information, or untracked documents. The solution isn’t glamorous, but it is essential.

Establishing a data contract framework is vital for managing structured data from sources like ERP, CRM, and HRMS, which should publish validated, versioned schemas. Regular data quality checks ensure completeness, timeliness, and accuracy. Clear ownership of golden data domains maintains integrity.

For unstructured data—documents, contracts, multimedia—a centralised metadata catalogue and lineage tracking facilitate classification by sensitivity and relevance, enhancing security.

These practices significantly reduce reconciliation efforts, up to 40% in large enterprises.

Remember, flawed foundational data amplifies issues in AI layers, making quality essential for reliable outcomes.

2. GenAI Solutions — Go Beyond the Obvious

Most CTOs today are being pitched for enterprise licenses of large language models and AI coding assistants. These are very powerful tools, but uncontrolled adoption can quickly burn budgets.

Begin with measured pilots, tracking token usage, time saved, cost, and a measurable return on investment (ROI).

Develop relevant custom contextual utilities atop foundational models; for example, creating GenAI tools that utilise the company’s internal data and large language models within a Retrieval-Augmented Generation (RAG) architecture to generate content automatically.

This approach can significantly enhance enterprise productivity. For instance, project managers no longer need to manually create requirements or design documents from scratch, as these GenAI utilities can automatically generate project documents, reducing the effort from days to just a few hours.

To quantify ROI, consider a consulting organisation with 1,000 employees: a 15% time-saving across delivery teams would free up the equivalent of 50 full-time employees annually. This demonstrates the substantial efficiency gains achievable through the strategic implementation of GenAI.should not be a shiny tool; it should be a workforce multiplier. Build guardrails, measure outcomes, and reinvest the gains.

3. Train Your Workforce — Don’t Just License Tools

Giving untrained employees access to powerful AI tools is like handing out race cars without seatbelts or manuals.

Yes, LLMs are intuitive — but productivity only compounds when users are trained in prompt engineering, data sensitivity, and critical validation.

Every enterprise AI rollout must include structured enablement through role-based learning paths, such as “AI for Engineers” or “AI for Analysts.”

Sandbox environments provide safe spaces to experiment with real data without any compliance risk. Certification programs not only add credibility to training efforts but also foster healthy competition among peers, encouraging them to become subject-matter experts in AI relevant to their lines of business.

Besides, upskilling the workforce to utilise AI tools efficiently can significantly enhance employee productivity, potentially doubling it. In today’s digital landscape, AI fluency is no longer optional but has become the new digital literacy.

4. Build an Enterprise Knowledge Platform

Across industries, knowledge duplication is one of the largest hidden productivity drains. On average, employees spend 4–6 hours a week searching for documents, slides, or answers that already exist elsewhere. That’s roughly 8–10% of total productivity lost to content scavenging.

The fix: consolidate your organisational knowledge into a centralised enterprise knowledge platform and power it with natural language search using GenAI.

This means indexing every proposal, design document, policy, and research note — structured and unstructured — in a vector database, then enabling conversational retrieval via an enterprise LLM.

Employees should be able to ask, “Show me all supply chain load consolidation projects we did for beverage companies in North America in the last 3 years,” and get accurate, contextual answers — instantly.

Integrating role-based access control (RBAC) into the knowledge platform not only ensures safe data retrieval but also protects data privacy and user-level data security.

5. Make Security Non-Negotiable

AI productivity without security is a ticking time bomb. Every GenAI project must be vetted by your Information Security and CISO teams from day zero — not as an afterthought.

Critical considerations include data privacy, which mandates the use of enterprise instances of LLMs and the strict avoidance of exposing sensitive data to public APIs.

Access control is essential as well; implementing role-based access control (RBAC) to govern user roles during the retrieval and redaction of sensitive content helps safeguard information.

Additionally, ensuring auditability is vital—every AI output should have traceable lineage and citations to prevent black-box scenarios.

Compliance with industry governance frameworks such as GDPR, HIPAA, and SOC2 is also necessary.

Remember, security should not be viewed as a blocker but rather as an enabler of productivity. Employees are more confident in using tools when they trust their security measures.

6. Governance and Monitoring — Measure What Matters

Finally, no productivity improvement is truly meaningful until it is measured. To ensure accurate assessment, establish clear KPIs for each department.

For instance, in engineering, track story points delivered per sprint, reductions in code review time, or improvements in sprint velocity. In consulting and delivery, measure the reduction in document creation time or proposal turnaround time.

Finance and HR can be evaluated based on time saved in reconciliation, data entry, or report generation.

Overall Enterprise Productivity can be tracked with newer metrics like “AI productivity index” to measure the time saved per employee per month. Even a 5% improvement at enterprise scale is equivalent to weeks of extra productivity annually.

Governance isn’t about bureaucracy; it’s about demonstrating impact. AI isn’t the goal — it’s the accelerator. The real transformation happens when technology, data, and people come together under a unified strategy for measurable outcomes.

Because in the end, every CTO’s true north isn’t only AI adoption or automation — it’s sustainable, scalable, and data-driven Enterprise Productivity.

The post CTO’s Playbook to Enterprise Productivity appeared first on Analytics India Magazine.