Power
Rankine Sorbocharging
Fuel-specific efficiencies exceeding standalone cycle limits — proportional to available waste heat.
A sorbocharged power cycle designed around the AI factory.
Synterran is engineering an industrial symbiosis for AI infrastructure — profitable compute today, clean power tomorrow, and durable carbon products on the horizon. From one thermodynamic cycle. Waste heat from the co-located compute load feeds back into the cycle. Carbon capture is built into cycle closure, not added downstream. At full scale, the architecture extends upstream — disaggregating hydrocarbons pre-combustion, where the molecule still has the free energy to rearrange itself into durable carbon and electricity.
Approach
Power, compute, and carbon as one thermodynamic system.
Most power systems reject waste heat. Most AI factories treat it as a cooling burden. And most carbon capture is bolted onto power plants as downstream cleanup. Synterran designs all three as a single system, so no “waste” leaves without doing work first.
The Synterran Cycle is an internal hydrocarbon combustion sorbocharged Rankine architecture that converts waste heat — from the cycle itself and the co-located load — into chemical potential, recovering latent and sensible energy that conventional cycles reject and resetting the cycle for the next power stroke.
Carbon
Capture in cycle closure
Residual CO₂ is captured as part of how the cycle closes — the same way water is “captured” in a steam turbine.
Compute
Fuel-to-token integration
Electricity becomes saleable compute; waste heat returns to the cycle.
The Synterran Cycle
System architecture.
The system-level architecture connecting the power cycle to its operating context.
Development
Built in stages. Each pays its way.
Synterran is a profitable development-stage company based in Atlanta, operating AI compute infrastructure as part of its zeroth deployment. The next deployment brings the Synterran Cycle into the world retrofitted to an Allison 250 helicopter turbine, which will A/B co-power our next few racks of compute — a first major milestone in power generation.
Deployment 0
Compute operations
64× A100 GPU cluster at over a quarter-million annualized revenue, providing operational discipline and a real demand-side foundation.
Cycle development & IP
Architecture specification, component design, four provisional patent filings, and additional IP maintained as trade secrets.
Site, lab & team preparation
Facility, equipment, and team development for the Deployment 1 build program.
Deployment 1
1
2
3
Hardware provisioning & thermal integration
A rack-scale unit of liquid-cooled AI hardware on grid power, with waste-heat-driven desiccant enhancement bolted onto commodity cooling towers.
Self-generated power & inlet conditioning
Fully provisioned 8-rack compute node with Allison 250 turbines providing B-side power alongside the grid. Desiccant inlet air chilling delivers deeply chilled, dried air to the turbine through direct contact with waste-heat-driven sorbent infrastructure.
Cycle closure
Full Synterran Cycle integration including constitutive carbon capture at pilot scale. The cycle provides primary power to the compute node with the grid as backup.
Deployment 2+
Legacy gas turbine retrofit
Validated cycle performance applied to stranded and underutilized gas turbine infrastructure at scale.
HydroCarbon Forge development
Pre-combustion carbon extraction pathway. Concrete deployment plans under development.
Purpose-built SCGT
A gas turbine designed around the Synterran Cycle from the ground up, rather than retrofitted to an existing platform.
Founder
Brendan van der Es
I’m a physics guy with over a decade of deep technical work across a diverse set of domains, from refinery automation and real-time fraud models to enterprise-scale data and service platforms. Synterran applies the same integrative instinct to revenue-aligned industrial symbiosis.
LinkedIn →
Intellectual property
The moat.
Four provisional patents protect the cycle architecture, sorbent cooling systems, load-side thermal integration, and compute infrastructure design. Additional mechanisms are held as trade secrets until development begins. The IP portfolio spans the full thermodynamic stack — from power generation through carbon capture to the compute load itself. That stack creates an exclusive pathway to recommissioning stranded gas turbine infrastructure — more economical than greenfield alternatives, and cleaner than most green technologies through constitutive carbon negativity.
Invention areas
- A novel thermodynamic power cycle
- Sorbent-based cooling for high-density compute
- Modular, vertically integrated compute infrastructure
- Load-side thermal harvesting
AI Factory
Infrastructure for AI workloads.
Synterran operates a reservable A100 cluster and is seeking pre-commitments on Deployment 1 hardware. The current target is NVL72, but hardware specification is an open conversation with committed partners.
At this scale, every deployment is a direct partnership — hardware selection, network topology, and hands-on operational support.
Get in touch
Compute inquiries, partnership conversations, or general questions.
Long view
From cycle closure to carbon materials.
Hydrocarbons are the feedstock of economical carbon negativity.
The near-term architecture captures carbon as part of cycle closure. The broader architecture goes further: the HydroCarbon Forge disaggregates hydrocarbon fuel before combustion, so carbon exits as durable construction materials — into a market that consumes more than 6 Gt annually — while the remaining hydrogen-enriched energy drives the cycle.
At deployment scale, these two carbon pathways combine. The power cycle captures residual combustion carbon. The Forge extracts carbon before combustion even occurs. The result is an industrial symbiosis between AI compute and carbon materials production — carbon-negative by thermodynamic design, not by offset or pledge.
At 250 GW, three thresholds converge: global-scale compute capacity, structural carbon output sufficient to reshape the construction materials industry, and combined carbon sequestration and material displacement at a scale commensurate with global annual emissions — under conservative assumptions, not by offset or pledge. That’s the target. Preferably in my lifetime.