Inferize tech to address 'idle GPU tax' in inference
Company: Nebius
The claim, verbatim
Nebius CTO Danila Shtan said running inference well requires the whole system to respond when demand changes, and positioned the Inferize technology as built to close that gap and reduce what Nebius calls the idle GPU tax.
Source (primary)
Nebius Rallies 9% on an Inference Deal While Its Earnings Multiple Sits Near 197x; CoreWeave Gains 4%, Oracle Advances 3% - Yahoo Finance (-, news_article)
View cached copy (2026-10-07)Live source ↗
Quote: “"Running inference well takes more than fast GPUs and optimized models." In Shtan's framing, the whole system has to respond when demand changes, and that gap is what Inferize was built to close.”
How we checked this
Checked on October 9, 2026. The cited source supports part of this claim, but not all of it.
Shtan's direct quote and the article's summary of his framing support most of the claim. But the article attributes the 'idle GPU tax' term to Nebius in general, not to Shtan's statement about Inferize.
Confirmed in the source:
- Danila Shtan is Nebius's chief technology officer
- Shtan said running inference well takes more than fast GPUs and optimized models
- The article characterizes Shtan's framing as the whole system needing to respond when demand changes, a gap Inferize was built to close
- Nebius calls the cost of idle standby GPU hardware the 'idle GPU tax'
We could not confirm this from the cited source:
- That Shtan himself tied the Inferize technology to reducing the 'idle GPU tax' (the article attributes that term to Nebius generally, not to Shtan's remarks)
That does not mean it is false — only that this source does not establish it, as of the date above. If we find a source that settles it, we will re-check and update this page.
What we did: Read our cached copy of the publisher (https://finance.yahoo.com/markets/stocks/articles/nebius-rallies-9-inference-deal-175729785.html) in full (12,488 characters, retrieved October 7, 2026) and checked each assertion in the claim against it.
Additional evidence
confirms Nebius Rallies 9% on an Inference Deal While Its Earnings Multiple Sits Near 197x; CoreWeave Gains 4%, Oracle Advances 3% - Yahoo Finance
Quote: “"Running inference well takes more than fast GPUs and optimized models." In Shtan's framing, the whole system has to respond when demand changes, and that gap is what Inferize was built to close.”