Are We Evaluating All the Options for Liquid Cooling?
Updated: 2 days ago
Observations from the 2026 OCP Global Summit raise important questions about the
future of data center cooling
The data center industry is undergoing an extraordinary transformation.
Artificial intelligence, high-performance computing, and increasingly powerful
processors are driving unprecedented investment in liquid-cooling technologies.
Equipment manufacturers are developing increasingly sophisticated solutions to
manage higher heat loads and greater computing densities.
Yet an interesting question is emerging:
Is the industry evaluating the full range of available cooling technologies, or is much of
its attention concentrated on improving the systems it already knows?
That question became particularly relevant during the 2026 Open Compute Project
(OCP) Global Summit in San Jose, California.
What NewTec Observed at OCP
Representatives from NewTec Partners attended OCP to engage with equipment
manufacturers, technology developers, and other participants in the rapidly evolving
data center industry.
One observation stood out.
When participants discussed liquid cooling, the conversation was overwhelmingly
focused on direct-to-chip (DTC) systems.
The conference featured extensive demonstrations of cold plates, coolant distribution
systems, manifolds, specialized connectors, and increasingly sophisticated equipment
designed to circulate coolant directly to high-performance processors.
Immersion cooling, by comparison, had a relatively limited presence in the
demonstrations and technical conversations NewTec representatives encountered.
Even among professionals actively developing liquid-cooling technologies, practical
familiarity with immersion appeared less widespread than expected.
This observation raises an important distinction.
There is a difference between an industry that has thoroughly evaluated a technology
and rejected it, and one that has concentrated its resources on a more familiar
approach while alternatives continue to develop.
The limited visibility of immersion cooling at OCP does not necessarily indicate a lack of potential. It may indicate an opportunity for more comprehensive evaluation, education, and industry collaboration.
Why Direct-to-Chip Has Gained So Much Momentum
The industry's concentration on DTC is understandable.
DTC allows manufacturers to address rising processor heat loads while preserving much
of the established server and rack architecture. It builds upon existing manufacturing
processes, service practices, and infrastructure designs.
The technology has also attracted significant engineering investment.
At OCP, NewTec representatives observed highly advanced DTC arrangements, including systems incorporating extensive networks of coolant lines, cold plates, manifolds, quick-disconnect fittings, and leak-detection equipment.
The engineering behind these systems is impressive.
But the increasing sophistication of DTC also introduces an important question.
As the systems required to cool individual processors become more complex, should
the industry be evaluating fundamentally different cooling architectures alongside
continued DTC improvements?
This is not an argument that DTC is unreliable, or that its engineering is misguided.
It's a question about long-term optimization.
Industry Conversations Reveal Interest in Alternatives
Discussions at OCP reinforced another encouraging observation: some major industry
participants remain interested in understanding how cooling technologies may evolve
beyond today's established approaches.
During one conversation with a representative of a major hyperscale data center
operator, NewTec discussed immersion cooling and the broader implications of
dielectric-fluid lifecycle management.
The discussion raised an important question that deserves greater attention:
What happens to dielectric cooling fluid after years of service?
NewTec's proposed approach involves recovering used Inherent Fluids™, converting
suitable materials back into isobutylene-derived molecular building blocks, and
manufacturing new fluid from recovered feedstocks.
This introduces a different perspective on the long-term economics and resource
efficiency of dielectric cooling.
Rather than treating fluid solely as an operating consumable, recovery and regeneration
create the potential to retain value throughout the fluid's lifecycle.
The conversation reinforced the importance of looking beyond ini4al thermal
performance and considering how cooling fluids will be managed, maintained,
recovered, and potentially regenerated over 4me.
For organizations making substantial long-term infrastructure investments, these
considerations could become increasingly important.
The Complexity Question
Another striking observation from OCP was the sophistication of the advanced
computing equipment and cooling infrastructure on display.
Modern high-density DTC systems demonstrate how far the industry has progressed in
managing increasingly demanding processor requirements.
They also illustrate the substantial mechanical infrastructure required to remove heat
from individual processors.
A DTC architecture may incorporate numerous coolant connections, fittngs, manifolds,
cold plates, distribution lines, and supporting control systems.
Each component has a specific engineering function. Each must also be installed,
operated, maintained, and supported throughout the equipment's service life.
Immersion cooling approaches the problem differently.
In a single-phase immersion system, compatible electronic equipment is submerged in
an electrically insulating fluid. Heat is transferred directly into the surrounding fluid and
carried to an external heat exchanger.
This arrangement can eliminate much of the internal cold-plate plumbing and individual
coolant connections required by conventional DTC designs.
However, immersion introduces different considerations, including fluid containment,
component compatibility, equipment servicing, pumping, and long-term fluid
management.
The central question is not which architecture has the fewest visible components.
It's which architecture delivers the best combination of performance, reliability,
serviceability, and total cost over its operating life.
That question becomes increasingly important as the value and computing density of
data center equipment continue to rise.
Are We Comparing Complete Systems?
A fair evaluation of DTC and immersion cooling must go beyond comparing cooling
components or fluid prices.
The analysis should include the complete installed system, encompassing:
• Cooling hardware and supporting infrastructure.
• Pumps, manifolds, piping, connections, and heat exchangers.
• Fluid performance and long-term management.
• Energy consumption and facility heat rejection.
• Maintenance, reliability, and potential downtime.
• Equipment compatibility and manufacturer support.
• Water consumption and environmental considerations.
• Installation costs, operating costs, and useful life.
The results will not necessarily favor the same architecture in every application.
For existing data centers, DTC may provide an attractive path to higher-density
computing without fundamentally changing established server infrastructure.
For new facilities designed around advanced cooling technologies, immersion may offer
opportunities to simplify aspects of the architecture and rethink how equipment and
supporting infrastructure are configured.
What matters is establishing these differences through credible engineering analysis and independent testing.
An Opportunity to Rethink Water and Energy Consumption
One additional observation from OCP was the relatively limited attention given to facility
water consumption in the discussions NewTec representatives encountered.
Much of the emphasis centered on removing heat from increasingly powerful
processors.
But removing heat from a processor is only the first part of the challenge.
That heat must ultimately be rejected from the facility.
Both DTC and immersion systems can employ evaporative, dry, or hybrid heat-rejection
approaches.
The temperature at which the cooling system operates can significantly affect the
efficiency and practicality of those arrangements.
Higher operating temperatures can make dry cooling more feasible, potentially reducing
water consumption and associated infrastructure requirements.
Purpose-engineered dielectric fluids may provide opportunities to develop immersion
systems capable of operating effectively at elevated temperatures.
The actual advantages must be demonstrated through testing and complete-system
engineering comparisons.
This raises a broader question:
Should cooling architectures be evaluated primarily on how effectively they remove
heat from processors, or on how efficiently and economically they manage that heat
from the processor all the way to the surrounding environment?
The latter may reveal opportunities that are less obvious when cooling equipment is
evaluated in isolation.
What Will It Take to Advance Immersion Cooling?
The OCP discussions reinforced another important point: broader immersion adoption
will require more than compelling technology.
Equipment manufacturers and data center operators need confidence in fluid
performance, materials compatibility, reliability, maintenance requirements,
qualification standards, and long-term supply.
Independent research and testing will be essential.
So will collaboration among fluid developers, equipment manufacturers, research
institutions, hyperscale operators, and industry organizations.
OCP is already advancing important work around immersion-fluid specifications,
materials compatibility, equipment requirements, and lifecycle management.
NewTec believes the next step should include a stronger emphasis on the role of fluid
chemistry itself.
Through Inherent Fluids™, NewTec is developing purpose-engineered isobutylene-
derived hydrocarbon fluids with molecular characteristics tailored toward demanding
dielectric cooling applications.
Rather than simply adapting existing base oils for immersion use, this approach offers
opportunities to explore how fluid properties and cooling-system design can be
optimized together.
That includes viscosity, thermal stability, dielectric performance, materials compatibility,
operating temperature, and fluid lifecycle management.
The objective isn't merely to develop another immersion fluid.
It's to investigate how better fluid engineering could help enable better cooling
systems.
Moving the Industry Conversation Forward
One of the encouraging lessons from OCP is that there appears to be an opportunity for
deeper technical discussion about immersion cooling.
DTC is advancing rapidly and will undoubtedly remain important to the industry's future.
But the continued advancement of DTC should not discourage rigorous evaluation of
other technologies.
As computing demands increase, the industry should remain open to solutions that
might reduce mechanical complexity, improve lifecycle economics, conserve resources,
or enable more efficient facility designs.
Immersion cooling does not have to replace DTC in every application to become an
important part of next-generation data center infrastructure.
It simply needs to demonstrate where it offers meaningful advantages.
That requires collaboration, qualification, independent testing, and a willingness to
examine the complete system rather than individual components.
Perhaps the most important takeaway from OCP is not that immersion cooling
received less attention than DTC. It's that the industry may still have important
questions to answer about what the best long-term cooling architectures should look
like.
At NewTec Partners, we believe those questions are worth exploring.
We welcome engagement with equipment manufacturers, researchers, data center
operators, and industry organizations interested in helping advance that understanding.
Comments