Kandou AI Inc., a developer of copper interconnects for artificial intelligence chips, has raised $225 million in funding.
Bloomberg reported today that the investment values the company at $400 million. The round was supported by Synopsys Inc. and Cadence Design Systems Inc., two publicly traded providers of chip development software. They were joined by semiconductor designer AIchip Technologies Ltd., SoftBank Group PBC and semiconductor-focused fund Maverick Silicon.
Graphics processing units in an AI cluster regularly exchange data with each other to coordinate their work. Usually, GPUs transmit traffic over fiber optic cables. Kandou AI is developing a copper-based alternative that it says is considerably more cost-effective.
The company also offers another copper-based interconnect, Glasswing, which can be used to connect components on a chip. Many modern processors include multiple compute and memory modules, each implemented on a separate chip. According to Kandou AI, Glasswing can move data between these modules faster and using less memory than competing technologies.
Glasswing is powered by a data transmission technology called CNRZ-5 Chord. It implements a proprietary variation of a networking approach known as differential signaling.
Differential signaling works by encoding each piece of data into two electrical signals with different voltages. These electrical signals travel to the target system via 2 separate wires. The system decodes the data contained in the signals by measuring the difference between their voltages.
CNRZ-5 Chord, the technology behind Glasswing, uses six wires to transmit data instead of the usual two. It can send up to five bits at a time over these links using half the power required by competing technologies. Additionally, Kandou AI claims that CRNZ-5 can transmit data over longer distances, giving chip designers more flexibility in how they design processors.
The company says its technology helps shrink the core chipset of a multi-chip processor. This also removes the need for an interposer, a component often used to move data between chiplets. These changes reduce the complexity of processor design, making them easier to produce and thus reducing manufacturing defects.
Kandou AI reportedly began prioritizing the AI infrastructure market after the arrival of CEO Srujan Linga last year. The Swiss company had previously focused on manufacturing components for connected devices such as industrial computers. The products it offers on this market help improve the reliability of USB and PCIe connections. Additionally, it sells a switch that connected devices can use to exchange data with peripherals.
The company offers its data center and consumer products as well as an interconnect diagnostic application. Engineers can use the software to test the reliability of network links they build using the company’s hardware.
Kandou AI is the latest in a series of interconnected startups to have raised funding since the start of the year. Xscape Photonique Inc. a startup that develops laser interconnects for AI clusters, farm a $37 million funding round earlier this month. A few days earlier, Ayar Labs Inc. raised $500 million from a consortium including several major chipmakers.
Photo: Unsplash
Support our mission of keeping content open and free by engaging with the CUBE community. Join the trusted network of CUBE alumniwhere technology leaders connect, share information and create opportunities.
- Over 15 million viewers of theCUBE videosfueling conversations about AI, cloud, cybersecurity and more
- More than 11.4,000 CUBE alumni — Connect with 11,400+ technology and business leaders shaping the future through a unique network of trust.
About SiliconANGLE media
Founded by technology visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a vibrant ecosystem of industry-leading digital media brands that reach more than 15 million elite technology professionals. Our new proprietary theCUBE AI Video Cloud innovates audience engagement, leveraging the CUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.
