Side-by-side comparison of AI visibility scores, market position, and capabilities
Two-phase immersion cooling using nuclear reactor-inspired subcooled nucleate boiling for AI data centers; $1M revenue with first 4MW contract delivered competing with GRC and LiquidStack.
Ferveret is a data center cooling technology company developing two-phase immersion cooling systems based on subcooled nucleate boiling — a heat transfer mechanism inspired by nuclear reactor cooling methods that achieves extremely high heat flux removal capability, enabling cooling of next-generation AI accelerators and high-performance computing chips that air cooling and single-phase liquid cooling cannot adequately handle. Founded in 2021, Ferveret raised $2.1 million from Y Combinator and E14 Fund, achieving $1 million in revenue in 2024 and successfully delivering its first 4 MW cooling contract from its El Paso, Texas manufacturing facility.\n\nFerveret's two-phase immersion approach works by submerging computing hardware in a dielectric fluid — when the chips generate heat, the fluid boils at precisely controlled temperatures, carrying heat away as vapor (the phase change enables far more heat transfer than single-phase liquid cooling). The subcooled nucleate boiling technology optimizes the boiling conditions for maximum heat transfer efficiency at controlled temperatures, enabling cooling of 300-1000W+ per chip that modern AI training accelerators (H100, B200) require. This approach addresses the fundamental limit that air cooling reaches at approximately 50W/chip.\n\nIn 2025, Ferveret competes in the data center thermal management market with GRC (Green Revolution Cooling, immersion cooling leader), LiquidStack, Submer, and traditional CRAC/CRAH air cooling for high-density AI compute installations. The data center cooling market has grown dramatically as AI training and inference workloads drive GPU density requirements beyond what air-cooled facilities can handle — NVIDIA H100 and B200 cards require 700W-1000W each, and the data centers being built for AI in 2024-2026 are designed for 40-80kW per rack, impossible with air. The successful 4MW delivery validates Ferveret's manufacturing capability. The 2025 strategy focuses on growing AI data center contracts with hyperscalers and colocation providers, scaling manufacturing capacity, and improving system density and heat reuse efficiency.
Cambridge/Colorado trapped-ion quantum computing (Honeywell majority; $625M+/$5B valuation Jun 2024); Helios Nov 2025 at 98 physical/48 logical qubits with 99.9975% fidelity serving Amgen/BMW/JPMorgan competing with IBM Quantum.
Quantinuum is a Cambridge, UK and Broomfield, Colorado-based integrated quantum computing company — majority owned by Honeywell (NASDAQ: HON) with $625+ million in total funding including a $300 million round led by JPMorgan Chase at a $5 billion valuation in June 2024 — operating the world's most accurate commercial quantum computers using trapped-ion technology combined with quantum software from Cambridge Quantum. In November 2025, Quantinuum launched Helios, its third-generation quantum computer featuring 98 physical qubits and 48 logical error-corrected qubits with 99.9975% single-qubit gate fidelity and 99.921% two-qubit gate fidelity — the highest-accuracy general-purpose commercial quantum computer commercially available. Serving enterprise customers including Amgen (drug discovery), BMW Group (materials simulation), JPMorgan Chase (financial optimization), and SoftBank Corp. (AI acceleration), Quantinuum was formed in November 2021 through the merger of Honeywell Quantum Solutions and Cambridge Quantum Computing. CEO Ilyas Khan.
Monitor how your brand performs across ChatGPT, Gemini, Perplexity, Claude, and Grok daily.