The software isn’t the first noteworthy aspect of the Tata-OpenAI data center agreement. The electricity is the cause.
Last week, as engineers and executives gathered at New Delhi’s Bharat Mandapam under glowing blue AI logos, the topic of megawatts kept coming up. Not algorithms. Not graphics processing units. Megawatts. There was a silent realization that this wasn’t just another tech collaboration as engineers pointed to schematics of liquid-cooled racks. Something more tangible was involved. heavier.
| Category | Details |
|---|---|
| Companies | Tata Group, Tata Consultancy Services (TCS), OpenAI |
| Founded | Tata Group: 1868; OpenAI: 2015 |
| Headquarters | Tata Group: Mumbai, India; OpenAI: San Francisco, USA |
| Core Businesses | Tata: Conglomerate spanning IT, energy, manufacturing; OpenAI: Artificial intelligence research and deployment |
| Data Center Project | Initial 100 megawatts capacity, scalable to 1 gigawatt |
| Initiative Name | Part of OpenAI’s global Stargate infrastructure program |
| Strategic Goal | Build AI-ready infrastructure and deploy ChatGPT Enterprise across Tata |
| India ChatGPT Users | Over 100 million weekly users |
| Energy Source | Powered significantly by green energy infrastructure |
| Reference | https://www.reuters.com/business |
It feels more like a shift in gravity than a business deal for OpenAI to become the first client of Tata Consultancy Services’ new HyperVault data center business, which will start with 100 megawatts and could grow to 1 gigawatt. Investors appear to think that chips alone are no longer the true scarcity in AI. It’s strength.
Given that the tech industry has been fixated on silicon for decades, that is an odd notion.
Naturally, Tata has always felt at ease handling heavy objects. Steel. railroads. automobiles. power plants. There is still a sense of scale as you pass Tata’s older industrial campuses in Jamshedpur—large chimneys, silently humming conveyor belts, and the odor of heated metal. These instincts are now being refocused on energy computation, which is an invisible but no less demanding task.
A portion of the story is revealed by the numbers. According to conventional standards, a 100 megawatt AI data center is massive. It would rank among the biggest AI infrastructure hubs in the world if it were scaled to 1 gigawatt. Although it’s still unclear if demand will increase quickly enough to immediately fill that capacity, OpenAI seems to be placing a wager that it will.
OpenAI seems to be more than just a space rental company. It’s protecting land.
This collaboration is also part of the much bigger Stargate project, which is OpenAI’s worldwide infrastructure endeavor that is purportedly supported by hundreds of billions of dollars. The tone of the ambition is almost industrial—more Hoover Dam, less Silicon Valley garage.
Once more, energy is turning into fate.
Simple physics contributes to this urgency. Large AI models require astronomical amounts of electricity to train and operate—sometimes using as much energy as small towns. Rows of GPUs inside contemporary AI facilities produce heat so intense that conventional air cooling cannot handle it. Like what Tata intends to implement, liquid cooling seems more like a need than an improvement.
There is a startling realization that AI is turning into a physical object when viewing footage of these systems, which are metal tubes that move coolant through server racks. Something that takes up room, uses resources, and creates a trace.
Interestingly, India has taken center stage in this narrative.
The nation is one of OpenAI’s fastest-growing markets, with over 100 million users using ChatGPT each week. However, usage by itself cannot account for the urgency. AI companies are being quietly forced to build locally due to data sovereignty regulations, latency requirements, and security concerns.
It is no longer sufficient to run AI from distant servers.
Here, Tata’s involvement seems all but inevitable. It has a certain institutional gravity because it is the most illustrious conglomerate in India. Workers entering Mumbai’s Tata Consultancy Services offices frequently see framed pictures of aviation pioneer J.R.D. Tata gazing serenely off into the distance. A sense of continuity prevails. Reimagining without losing one’s identity.
Tata is now constructing computational infrastructure in place of airplanes.
The implementation of ChatGPT Enterprise throughout Tata’s workforce has a subtly strategic element as well. Hundreds of thousands of engineers will start writing code, analyzing data, automating procedures, and incorporating AI tools into their everyday workflows. This internal change might be just as significant as the data center itself.
Behavior is altered by infrastructure.
Similar actions are taking place around the world. Billions of dollars are being invested in AI data centers by Google, Microsoft, Amazon, and Meta. Tech companies are suddenly courting energy companies as well. Utility companies and technology companies are becoming less distinct from one another.
Silicon Valley is becoming aware of its reliance on electrical grids.
It’s difficult to ignore the similarities to past industrial revolutions. Manufacturing was changed by electricity at the beginning of the 20th century. Power sources became closer to factories. Infrastructure became the foundation of entire cities. AI appears to be causing a similar reorganization.
However, intelligence is now produced by the factories.
Additionally, there is a geopolitical component. AI infrastructure is becoming more and more important to nations. Local compute hosting provides influence, security, and control. India seems keen to draw in these investments as it positions itself as a center for artificial intelligence.
That goal is perfectly aligned with Tata’s collaboration with OpenAI.
But there are still unanswered questions. The demand for AI has grown quickly, but infrastructure investments are costly and take time. Reversing the construction of a gigawatt-scale facility is not an easy task. Investors appear upbeat, but tech has been duped by optimism in the past.
Silently, memories of previous cycles of overbuilding persist.
The ultimate bottleneck might turn out to be energy itself. Even green power, which Tata intends to use extensively, necessitates extensive planning and funding. Creating electricity is just one aspect of the problem. It is quite another to deliver it consistently, reliably, and economically.
Outages are not well tolerated by AI. There is an odd feeling of familiarity as you watch this play out. Once characterized by weightlessness—code floating in the cloud—the tech sector is now gaining weight once more. rooted. reliant on tangible infrastructure. Pure software is losing its allure.
And Tata, a business that was founded during the era of steel mills and railroads, is now unexpectedly at the forefront of AI’s future. Not due to the genius of software. But because it understands how to create things that use energy, withstand strain, and endure for decades.
