No lab has a once-and-for-all fix — but three point the way out of the noise

Follow-up question: which academic labs are doing research here and potentially have the best solution to solve data-center community noise once and for all?

Evidence: 116 live-web research records gathered 2026-08-29 — 20 on EPFL's plasma absorber, a focused Boston University extraction, 7 passive-metamaterial papers, 60 active-control candidates and 27 advanced-cooling programs — distilled to 9 distinct academic groups. Many records restate the same study; they are not independent validations.

The research frontier: what each lab treats, and how proven it is

acoustic treatment pathcooling / source-elimination pathring marks the three shortlisted leadershover a lab for its claim, evidence badges and source

The shortlist

1 · high-upside acoustic moonshot

EPFL Acoustic Group, Signal Processing Laboratory

Hervé Lissek, Stanislav Sergeev, Romain Fleury. Ionizes a thin air layer into an active "plasmacoustic metalayer": experimentally demonstrated 100% absorption in its test condition, tunable from several Hz to the kHz range — a 20 Hz absorber only 17 mm thick where passive walls need ~4 m. Licensed to Sonexos.

Caveat: laboratory/component scale only — durability, ozone and electrical safety, large-aperture scaling and lifecycle cost unproven outdoors.

direct acoustic measurementcommercial partner

actu.epfl.ch · eurekalert.org

2 · strongest passive contender

Boston University — Xin Zhang research group

Xin Zhang and Reza Ghaffarivardavagh built an acoustic metamaterial device that blocked up to 94% of transmitted sound energy while preserving airflow — the key property for fan and duct openings, avoiding the pressure-drop penalty of barriers.

Caveat: lab prototype; 94% of sound energy is not 94% of perceived loudness, and tuned structures need testing under full-scale flow, changing fan speeds and multi-tone spectra. No data-center field pilot found.

direct acoustic measurementairflow test

bu.edu

3 · strongest source-elimination path

NSF ES2 consortium — UT Arlington lead

Binghamton, UT Arlington and Villanova run dedicated data-center, immersion-cooling and liquid-cooling labs. UTA's COOLERCHIPS work, led by Dereje Agonafer with partner Accelsius, brings two-phase direct-to-chip cooling that can remove or slow server fans — attacking a source instead of masking it.

Caveat: programs optimize heat, energy and density; no exterior community-noise measurement found. Heat still leaves the building — pumps and heat rejection remain.

data-center testbedcommercial partner

eurekalert.org · uta.edu

What the map cannot say alone

The top-right of the frontier is empty: no group pairs source elimination with field-verified community acoustics — even EPFL's "perfect absorption" is a chamber result, not a property-line test. actu.epfl.ch

The 20 EPFL records largely restate one Nature Communications study (Sergeev, Fleury & Lissek, 2023) and its Sonexos partnership — corroboration, not twenty validations. sciencedaily.com

Cooling programs' noise benefit is inferred source reduction: UTA's COOLERCHIPS partnership targets cooling that can consume up to 40% of a data center's energy, but publishes no exterior sound data. dallasinnovates.com

The watchlist's most data-center-relevant acoustic test is Shanghai Jiao Tong's over-the-rotor treatment of a ducted fan in an anechoic chamber — tonal fan noise is exactly the community complaint. doi.org

The missing proof: a 12–24 month integrated full-scale pilot

Eliminate at the sourceDirect-to-chip / immersion cooling (ES2 · UTA style) removes or slows server fans and shrinks air-moving demand.

→

Treat residual airflowEPFL/Sonexos active plasma treatment or BU-style passive metamaterial on remaining heat-rejection ducts and fan outlets; isolate transformers, chillers, pumps, generators.

→

Verify at homesIndependent measurement at source, property line and residences: dBA plus octave/third-octave, tonal and low-frequency metrics.

→

Stress the systemNormal operation, hottest design day, night low ambient, partial and full load, equipment synchronization, generator tests.

Verdict

Best near-term route to quieter campuses: ES2/UTA-style liquid or immersion cooling plus mature source isolation and careful siting.
Best high-upside technology for the stubborn low-frequency hum: EPFL's plasmacoustic metalayer / Sonexos pathway.
Best passive contender: Boston University's airflow-preserving acoustic metamaterial.
Best overall answer: an integrated system, not one invention — even perfect fan silencing would not silence transformers, chillers, pumps or generators. A successful field pilot, not another chamber test, is the missing evidence.

All nine groups

Group	Approach	Reported evidence	Badges	Source

Live-web research pass dated 2026-08-29: 116 extracted records across five result sets (EPFL plasma coverage, BU extraction, passive-metamaterial papers, active-control candidates, advanced-cooling programs), reduced to 9 distinct academic groups. Frontier positions are qualitative editorial judgments on two axes — treats noise after creation vs eliminates the source, and lab concept vs field/commercial readiness — not measured scores; badges mark evidence explicitly present in the sources. Duplicate press coverage of the same studies and non-academic vendors were cut for space. "94% sound energy" and "100% absorption" are laboratory claims under test conditions, not field performance.