About
Rigorous engineering knowledge for the people building AI infrastructure.
Drybulb is a resource hub that publishes deep technical writing on the engineering of AI factories and mission-critical data centers. The site exists because the best knowledge in this field is still locked inside organizations, shared only in conference hallways and vendor-filtered whitepapers. We think it should be public.
Every article on Drybulb is grounded in public-domain knowledge and first-principles engineering. We cover the full infrastructure stack — power, cooling, networking, reliability, and sustainability — with the rigor that practitioners expect and the depth that marketing collateral can't provide. Alongside the writing, we're building a growing library of free, open tools for the engineers who design and operate these facilities.
Who this is for
Owners & Developers
Building or expanding data center capacity and navigating the engineering decisions that determine whether a facility delivers what the program requires — from cooling strategy to power redundancy to phasing logic.
Investors & Lenders
Evaluating data center and AI infrastructure assets and need to close the gap between claimed capacity and deliverable capacity before committing capital.
Engineering & Design Teams
Working across power, cooling, and MEP disciplines on high-density AI facilities and looking for independent technical depth on the coordination problems that cross disciplinary boundaries.
Program & Operations Leaders
Planning early-stage AI clusters or hyperscale campuses and need to get site criteria, utility intake, density assumptions, and phasing logic right before design begins.
Editorial standards
Technically rigorous
Every hard figure is verified against manufacturer specs, industry standards, or first-principles calculation. We show our reasoning.
Public-domain knowledge only
Nothing on Drybulb requires insider access, NDA-protected data, or proprietary information. Every claim can be independently verified by the reader.
Independent and unsponsored
Drybulb does not accept sponsored content. Equipment and technologies are evaluated on engineering merit, not commercial relationships.
What we cover
Power Systems
- —MV/HV utility intake and substation design
- —Distribution, transformation, and redundancy strategy
- —UPS architecture and BESS integration
- —Grid-constraint planning and demand response
Cooling & Thermal
- —Direct-to-chip liquid cooling (cold plate / CDU)
- —Immersion cooling (single-phase and two-phase)
- —Facility water systems and cooling towers
- —PUE / WUE optimization and heat reuse
AI Infrastructure
- —AI factory and hyperscale campus design
- —High-density GPU cluster layout and planning
- —Scale-up and scale-out network fabrics
- —Rack power density strategy and roadmapping
Reliability & Sustainability
- —Redundancy tiers and concurrent maintainability
- —Commissioning planning and execution
- —Life-cycle carbon and embodied-energy analysis
- —Water stewardship and waste-heat recovery
About the editor
Drybulb was started by an independent engineer — a licensed Professional Engineer (PE) and Doctor of Design — with 25+ years of experience across the full data center infrastructure stack. The motivation was simple: the best engineering knowledge in this industry is hard to find unless you already know the right people. This site is an attempt to change that, one article and one tool at a time.
Start with the cornerstone article.
A deep engineering overview of how AI factories differ from traditional data centers.