Kya hua
Intel spinoff Cornelis ne IAG Capital ke lead me $205 million raise kiye aur Active Compute Fabric platform unveil kiya jo AI chips ke beech fast communication deta hai.
AI infrastructure startup Cornelis ne $205 million ka funding round raise kiya hai, lead IAG Capital Partners ne kiya. Company AI chips ke beech communication fast karne ke liye fabric banati hai taaki GPUs compute aur data transfer ek saath kar saken.
Cornelis 2020 me Intel se spin-off hua tha aur Nvidia ke closed stack ka open alternative ban raha hai. Nvidia GPUs dusre fabrics par chal sakte hain par unka software Nvidia stack par hi best chalta hai.
Naya platform Active Compute Fabric is pitch ka centre hai. CN5000 400 Gbps par already shipping hai, jabki CN6000 800 Gbps aur multi-protocol support ke saath Q4 2026 me ayega. Dono Ultra Ethernet aur UALink support karte hain.
Yeh funding badi trend ka hissa hai jisme Nexthop AI ($500M) aur Eridu ($230M) bhi Nvidia dependence kam karne ke liye paisa utha rahe hain. Cornelis ka estimate hai ki 100K-GPU cluster me aadha time waste hota hai — ~$1.68B aur 500 GWh saal ka.
Cornelis yahan tak kaise pahuncha
Intel se nikal kar open AI fabric challenger banne tak ka safar.
- 2020 — Cornelis Intel se spin-off hua high-performance networking par focus ke liye.
- 2026-09-14 — $205M round aur Active Compute Fabric ka announcement hua.
- Abhi — CN5000 400 Gbps par shipping; CN6000 800 Gbps Q4 2026 tak target.
- Aage — Qualcomm partnership ka showcase AI Infra Summit par plan hai.
Aage kya dekhna hai
Q4 2026 me CN6000 ki delivery aur Qualcomm showcase batayega ki open standards Nvidia-locked customers ko convert kar pate hain ya nahi. Real performance aur pricing hi adoption decide karega.
Kyun matter karta hai
AI data centers ko Nvidia lock-in se bahar ek fast, open switch milega — CN5000 400 Gbps par shipping, CN6000 800 Gbps par Q4 2026 me.
Money
Business Executives · Developers
Data-center operators billions bachaa sakte hain — Cornelis ke model me 100K-GPU cluster me chips ke idle rehne se ~$1.68B/year waste hota hai.
Work
Developers · Business Executives
AI cluster banane wali teams ko vendor-neutral option milega jo kisi bhi GPU/accelerator par chale, Nvidia lock-in kam hoga.
Daily Life
General Public
Zyada efficient fabric se bijli ki badi bachat ho sakti hai — 100K-GPU cluster par ~500 GWh, jab AI ke power use par sawal uth rahe hain.
Seedhi baat
Open networking se billions ka GPU waste bach sakta hai.
Cornelis ne $205M uthaye — 800 Gbps open fabric se Nvidia ko challenge.
Verified sources (4)
AI infrastructure company Cornelis raises $205M to chip away at Nvidia's dominance
↗Cornelis raises $205M to challenge Nvidia's dominance in AI networking
↗Cornelis Raises $205M for AI Networking, Unveils Active Compute Fabric with Qualcomm
↗Cornelis raises $205 million to challenge Nvidia with open AI networking
↗Claims and linked sources
7 claimsAI infrastructure company Cornelis ne IAG Capital Partners ke lead me $205 million ka round raise kiya hai taaki AI chips ke beech communication fast ho.
Cornelis ne Active Compute Fabric launch kiya hai jo chips ko compute aur data transfer ek saath karne deta hai taaki GPUs ka wait time kam ho.
Cornelis 2020 me Intel se spin-off hua tha aur open architecture deta hai jo kisi bhi GPU/accelerator par chalta hai, ek vendor se bandha nahi.
Nvidia GPUs technically dusre fabrics par chal sakte hain par Nvidia ke apne software par best optimize hain, jisse customers usi stack me rehte hain.
Cornelis ka CN5000 switch 400 Gbps par already shipping hai, jabki CN6000 800 Gbps aur multiple protocols ke saath Q4 2026 me broadly available hoga.
Cornelis Ultra Ethernet aur UALink jaise open standards support karta hai jo Nvidia ke proprietary stack ke alternative ke liye bane hain.
Cornelis ne Qualcomm Technologies ke saath strategic partnership announce kiya hai jo AI Infra Summit par showcase hoga.