Vancouver Startup Wafr Secures $100 Million to Cut AI Data Centre Water Use by 95 Percent

Wafr Technologies, a Vancouver-based startup founded in 2025, announced Thursday that it has raised $100 million from private investors to commercialize a cooling technology designed to slash water and energy use at artificial intelligence data centres. The company plans to deploy the capital toward launching an AI research lab, establishing a new data centre facility, and hiring dozens of technical staff across Canada over the coming months.

The timing reflects intensifying pressure on the AI infrastructure sector to address a critical sustainability bottleneck. A single 100-megawatt AI data centre in the United States consumes roughly 2 million litres of water daily, equivalent to the water use of 6,500 households, according to a 2025 analysis from the International Energy Agency. As large hyperscalers including Amazon, Google, and Microsoft scale up computational capacity to support generative AI workloads, the environmental footprint of cooling these facilities has become a licensing and regulatory concern across North America and Europe.

Wafr’s proprietary thermal battery system captures and stores cooling capacity when grid electricity is inexpensive, then releases it during peak demand periods. Co-founder Darrell Kopke, who also serves as Vancouver executive director for the Creative Destruction Lab, stated the technology can reduce water use by up to 95 percent and cooling power consumption by up to 80 percent compared to conventional evaporative cooling systems.

Market Concentration and Capital Plans

The startup has already begun commercialization and signed letters of intent with international partners constructing AI data centres, though Kopke declined to name the investors in the initial funding round. Wafr intends to raise an additional $200 million “as soon as possible” from government funds and private investors to accelerate rollout.

The company is prioritizing the United States market alongside European nations such as Germany. The strategic focus reflects data centre density: the U.S. operates approximately 5,300 data centres, while Germany, the next-largest declared hub, operates around 400. Kopke emphasized that the U.S. also represents the market with the “least sustainable AI” infrastructure, presenting both an opportunity and a business imperative for efficiency gains.

Wafr’s approach sits within a broader wave of climate-focused infrastructure funding. As AI inference systems and computational efficiency have become funding priorities, solutions targeting the power and water constraints of data centre operations have attracted institutional interest.

The Cooling Bottleneck Problem

Hyperscale AI data centres house densely packed advanced computing chips that generate significant heat during operation. Traditional cooling relies on water-intensive evaporative systems, tower cooling, or direct liquid cooling loops. For operators managing facilities at 100 megawatts and above, cooling costs and water availability have become material constraints on expansion capacity, particularly in water-stressed regions.

Kopke characterized the challenge plainly: “This crazy, massive market has a major bottleneck, which is power and water. Our technology virtually eliminates the need for water and stops evaporative water for cooling.”

The competitive and regulatory environment underscores the urgency. Data centre operators face mounting pressure from state and local governments to demonstrate environmental stewardship, particularly in water-constrained jurisdictions. Simultaneously, major cloud providers have committed to net-zero emissions targets, making energy and water efficiency a core technical problem rather than an optional enhancement.

Capital Momentum in AI Infrastructure

Wafr’s $100 million raise reflects confidence in the addressable market for AI infrastructure efficiency. The company’s stated intent to raise a further $200 million suggests backers view the commercialization timeline as near-term, with existing letters of intent providing early validation.

The competitive landscape for AI data centre solutions extends beyond cooling. Other startups are pursuing liquid cooling, direct-to-chip approaches, and modular data centre designs to address similar constraints. Wafr’s thermal battery angle differentiates on grid arbitrage and peak-shaving economics, reducing both water and peak power demand simultaneously.

The funding round also signals confidence in Canada as a jurisdiction for advanced infrastructure development. Vancouver has emerged as a centre for AI policy innovation and computational research, with the Creative Destruction Lab serving as an accelerator for climate-tech and AI infrastructure ventures. Wafr’s expansion plans across Canadian regions reflect both domestic market opportunity and a talent recruitment advantage in the region.

What Remains Uncertain

Several questions remain unaddressed at this stage. The startup has not disclosed the specific deployment timelines for its first commercial installations or provided independent verification of the claimed efficiency gains across different data centre configurations and climate conditions. Scaling thermal battery technology from lab prototypes to production systems at hyperscale facilities presents engineering, reliability, and regulatory approval hurdles that remain ahead.

Additionally, the capital intensity of building proprietary data centre infrastructure alongside commercializing cooling technology to external customers represents a dual operational burden that few hardware-focused startups have successfully managed simultaneously.

Wafr’s success will likely hinge on securing early high-profile pilot deployments with established hyperscalers, demonstrating measurable efficiency gains in production environments, and closing the additional $200 million funding round to support geographic expansion and engineering talent acquisition. The thermal battery concept is not novel, but applying it systematically to the AI data centre cooling problem at scale remains an open engineering and business validation challenge.