Inferize is an Israeli AI infrastructure startup that developed technology to reduce the time required to load AI models onto GPUs from tens of minutes to seconds, enabling inference capacity to respond more closely to real-time demand. The company was founded by Guy Bortnikov and CTO Lior Gorbonos, both previously part of the founding team at Granulate, an infrastructure optimization firm acquired by Intel in 2022 for approximately $650 million. Inferize had 17 employees in Tel Aviv and had never publicly disclosed its existence or funding before the acquisition.
Led by TLV Partners, with participation from additional investors not named in the article
The deal, valued at up to $150 million for a 17-person company that raised only $10 million and had been operating for less than nine months, illustrates how much AI infrastructure companies are willing to pay for technology that improves the economics of running inference workloads at scale. Inferize's technology addresses a growing bottleneck: as AI inference demand fluctuates rapidly, long GPU load times leave expensive hardware idle or force users to wait, a small inefficiency that compounds into significant cost at scale. The acquisition gives Nebius the technology for its Token Factory inference platform and signals that even early-stage, pre-revenue infrastructure tools targeting inference efficiency can command acquisition premiums well above typical seed-stage multiples.