Daikin Bets $108M on India to Solve AI's Heat Problem
Daikin Industries has approved a ~$108 million investment to establish its first global R&D hub outside Japan in Haryana, India, focused on advanced cooling solutions for AI data centers. The move reflects the growing thermal crisis in AI infrastructure, where high-density chips like NVIDIA's Blackwell require fundamentally new cooling approaches. For the broader industry, it signals that critical AI infrastructure innovation is increasingly being anchored in the Global South.
## Why This Matters for PMs If you're building or procuring AI infrastructure — even if you're not the one signing the data center contracts — Daikin's India bet forces a specific question: how much of your product's roadmap is implicitly dependent on cooling technology that doesn't fully exist yet? This isn't abstract. High-density GPU clusters are already running into thermal constraints that throttle performance. If your team is planning AI workloads that require sustained compute at scale — training runs, real-time inference, multi-modal pipelines — you need to be in conversations with your infrastructure counterparts about thermal headroom, not just GPU availability. The bottleneck in 2027 and 2028 may not be chips. It may be the ability to cool them. Concretely: if you're evaluating cloud providers or colocation partners for AI workloads, start asking them directly about their liquid cooling capacity and their roadmap for high-density racks. Not all facilities are equal, and the gap is widening. Providers who have locked in relationships with companies like Daikin — or who have invested in liquid cooling infrastructure — will have a meaningful performance and reliability advantage over those still running air-cooled legacy setups. The urgency here is real. Daikin's Haryana facility won't be operational until mid-2028. The cooling solutions being designed there are meant for the infrastructure wave that follows the current one. If your product depends on AI compute at scale, the thermal constraints of the next 18 to 24 months are a present-tense planning problem, not a future one.
Daikin approved a ~$108M investment to build its first R&D hub outside Japan in Haryana, India, completing by June 2028.
The Data Center Has a Fever
Picture a server room in 2024. Now picture that same room packed with NVIDIA Blackwell chips running inference on large language models around the clock. The temperature analogy isn't metaphorical — it's thermal physics.
These chips run so hot that the cooling systems designed for the previous generation of computing simply can't keep up. Someone has to build what comes next. Daikin, the Japanese HVAC giant that has quietly kept the world's buildings comfortable for nearly a century, has decided that someone is them — and that the place to build it is India.
On June 26, 2026, Daikin's Board of Directors approved the establishment of Daikin Research and Development India Private Limited in Haryana. The capital investment stands at INR 8 billion, approximately $96 million USD, with the broader investment plan — first outlined in April 2026 — totaling roughly $108 million. Construction of the new facility is scheduled for completion by June 2028.
For a company that has historically kept its most sensitive R&D work inside Japan, this is not a routine expansion. It is a statement about where the world's engineering talent lives, and where the world's cooling problems are most acutely felt.
Why India, Why Now
Daikin's choice of Haryana isn't arbitrary. The region sits within striking distance of Delhi's expanding technology corridor and has become a manufacturing anchor for the company's India operations. But the deeper logic is demographic and climatic.
India is simultaneously one of the hottest countries on earth and one of the fastest-growing data center markets — a combination that makes it the perfect stress-test environment for next-generation cooling technology. If you can engineer a chiller that works efficiently in a 45-degree Indian summer, you can probably make it work anywhere.
There's also a talent calculation embedded in this decision. India produces hundreds of thousands of engineering graduates annually, many of them specializing in thermal systems, fluid dynamics, and embedded software — exactly the disciplines required to design cooling solutions for high-density AI infrastructure. Daikin isn't just opening a satellite office; it's trying to plant roots in an ecosystem.
The hub will focus on software and product development for large-scale chillers and air-conditioning systems tailored for data centers, commercial facilities, and residential applications across what Daikin calls the Global South — Asia, Oceania, and beyond.
The Thermal Crisis Behind the Investment
To understand why a Japanese HVAC company is making its largest international R&D bet right now, you need to understand what AI chips are doing to the physics of computing. Think of traditional server cooling like air conditioning a house — manageable, predictable, well-understood. Now think of AI training clusters like trying to cool a steel foundry using the same window unit.
The heat density per square foot has increased by an order of magnitude as chips like NVIDIA's Hopper and Blackwell architectures pack more computational power into smaller spaces.
Liquid cooling has emerged as the most promising response — circulating coolant directly across chip surfaces rather than blowing air through racks. Daikin has been positioning for this moment. The company's acquisition of Chilldyne, a liquid-cooling pioneer, signals that Daikin understands the transition underway: the future of data center thermal management isn't just bigger air handlers, it's fundamentally different physics.
The India R&D hub will be where that next chapter gets written, combining Daikin's hardware manufacturing depth with software-defined cooling controls and locally manufactured chillers.
A Quiet Power Shift in Global Infrastructure
There's a larger story embedded in this announcement that goes beyond one company's capital allocation. For decades, the critical infrastructure underpinning global computing — the chips, the networking gear, the cooling systems — was designed in the United States, Japan, and Western Europe, then manufactured and deployed everywhere else. What Daikin is doing in Haryana is part of a broader redistribution: moving the act of invention, not just production, closer to the markets that will consume these technologies most intensely.
The Global South is not waiting for Silicon Valley to solve its infrastructure problems anymore. It is increasingly building the institutions — the labs, the manufacturing lines, the engineering cultures — to solve them itself. Daikin's $108 million bet is a vote of confidence in that shift.
By June 2028, when the Haryana facility comes online, the AI infrastructure buildout will be several years deeper into its current trajectory. The companies that figured out how to keep those systems cool — efficiently, affordably, at scale — will have locked in relationships and intellectual property that will be very hard to dislodge.
Frequently Asked Questions
AI chips like NVIDIA's Hopper and Blackwell architectures pack far more computational density into smaller physical spaces, generating heat loads that air cooling systems cannot efficiently manage. A high-density AI rack can require 50-100kW of cooling capacity compared to 5-10kW for a traditional server rack. Liquid cooling — circulating coolant directly across chip surfaces — is increasingly the only viable solution at this density.
The data center cooling market is contested by HVAC incumbents like Daikin and Vertiv, alongside liquid cooling specialists like Chilldyne (now Daikin), CoolIT Systems, and Asetek. By establishing an R&D hub in India, Daikin is betting that proximity to a high-growth market and deep engineering talent will accelerate product development cycles faster than centralized R&D in Japan. The $96 million capital investment for the subsidiary suggests this is a long-term competitive positioning move, not an incremental expansion.
Yes, particularly for products requiring sustained, high-density GPU compute. Thermal constraints are already causing performance throttling in some colocation environments, and the gap between cooling supply and AI compute demand is expected to widen through 2027-2028. PMs planning significant AI infrastructure dependencies should audit their cloud or colo providers' liquid cooling capacity and ask explicitly about high-density rack availability as part of vendor evaluation.
