ANVE: the private AI office, on the record. See what that means

ANVE

Private AI, answered · updated 2026-10-07

How noisy is an AI workstation in an office?
What has been measured, what has not, and where to put the machine.

Short answer. It depends on the cooling design and on how hard the machine works, and published measurements are few. ServeTheHome measured a Framework Desktop with a Ryzen AI Max+ 395 at 36 to 37 dBA idle and 39 to 41 dBA under load in its noise-floor studio, and called it one of the quieter systems it has tested. Cards with a higher power rating make more heat to remove: 145 W for an RTX PRO 4000 Blackwell, 575 W for a GeForce RTX 5090, and their product pages publish no sound levels. AI work keeps the chip busy while an answer is generated, so judge noise under load, not at idle.

01

What has been measured

The measured system drew 10 to 20 W idle and 140 to 170 W under load in the same review, so the fans have to move roughly ten times as much heat when it works. Phoronix also found that a Framework Desktop was quiet enough for a typical office even with an extra fan fitted and every fan at full speed. Two reviews of one product are not a rule for every AI machine.

SystemIdleUnder loadSource and method
Framework Desktop, Ryzen AI Max+ 395, 128 GB, Noctua fan option36 to 37 dBA39 to 41 dBAServeTheHome, noise-floor studio, November 2025
Same machine with an extra front fan and all fans at 99 to 100 per centNot measuredA clear, persistent sound that Phoronix called not as loud as larger desktopsPhoronix, qualitative
Apple Mac Studio, NVIDIA RTX PRO 4000 Blackwell, GeForce RTX 5090Not listedNot listedNo sound level on the vendor pages checked
02

Why AI work is noisier than office work

Word processing and email draw power in short bursts. Generating an answer, or transcribing a meeting in the background, can keep the chip busy for as long as the answer or the recording lasts, and the fans follow the heat. Judge a machine by its sound after ten minutes of work, not by its sound when idle.

03

What decides the noise

  • Power. A 145 W card or a 45 to 120 W chip has far less heat to remove than a 575 W card. AMD lets its chip’s power limit be set in that 45 to 120 W range, and a lower limit means less heat.
  • Cooling design. Larger fans turning slowly are generally quieter than small fast ones, and bigger heatsinks help.
  • Placement. A machine in a closed cupboard runs hotter, so its fans run faster. A ventilated room away from desks is quieter for people and cooler for the machine.
  • Dust and age. Blocked heatsinks make fans work harder over time.
04

Placing it in an office

Because models are used over the network, the machine does not have to sit near people. A small ventilated room, a store room or a locked cabinet with airflow works, with a wired network cable. If it must sit in the open, ask for a demo, run a long answer or a meeting transcript, and listen from your usual seat. A phone decibel app is only a rough guide.

ANVE is pre-release and has not published noise, size or power figures, because no sealed unit exists yet. Ask any vendor for a measured sound level under load, with the test method and the distance.

05

Questions people ask

Is dBA the same as dB?
dBA is a decibel scale weighted to how the human ear hears. ServeTheHome reports dBA, so its figures can be compared with other dBA readings, not with unweighted decibels.
Can the machine be made quieter?
Yes, in three ways: lower its power limit if the vendor allows it (AMD lists 45 to 120 W for the Ryzen AI Max+ 395), give it more space and airflow, and move it out of the open office.
Will an AI box be louder than a normal PC?
Under a sustained load it can be, because the chip may work for the whole length of an answer. At idle the measured Strix Halo system was at the studio’s noise floor of 36 to 37 dBA.
06

What is not real yet

No sealed ANVE unit exists; every image of the appliance on this site is a labelled render. The software runs today on a development machine. No customers yet, no certifications, no delivery date, and the final price is not announced.

Reservations are ₹999 in India or US$12 elsewhere, refundable in full until you confirm an order in writing (terms). Offices that cannot use cloud AI can apply to be one of ten design partners.