Kandou AI Raises $225M to Solve Data Movement Bottlenecks in AI Infrastructure
Pressure builds in places most people never look. Not on the screen, not in the model, but in the narrow lanes where data is supposed to move like a clean heartbeat and instead starts stuttering. That friction, quiet and expensive, is where Kandou AI decided to make its money. Kandou AI just pulled in $225M in a Series A that did not tiptoe in, it landed with intent. Maverick Silicon led the charge, backed by SoftBank Group Corp., Synopsys Inc., Cadence Design Systems Inc., and Alchip Technologies Limited. The valuation clocks in around $400M, which feels less like a price tag and more like a statement about where the bottlenecks really are.
This is not a story about chips. It is a story about traffic. Too many lanes, not enough flow, and workloads piling up like rush hour with no exit ramp. Kandou AI leans into copper like it still has secrets left to give. While everyone else stares at optics like it is the promised land, Kandou AI is squeezing more signal, more distance, more efficiency out of what is already under our feet.
Srujan Linga, CEO, stepped in in 2025 and got straight to business. No theater, just direction. Dr. Amin Shokrollahi, CTO, founder, stayed close to the math, which is exactly where he belongs. One runs the tempo, the other writes the physics. That combination tends to age well.
Over 28M chips already carry their technology, with more than 13M custom chiplets delivered across a growing customer base. This is not a lab experiment dressed up for a pitch deck. This is deployed, scaled, and now capitalized to move faster.
The real play sits inside the memory wall. Systems do not just need smarter models, they need cleaner highways between them. Training, inference, CXL memory, rack level connectivity, all of it breaks down when data stalls. Kandou AI is not trying to out-hype the market. It is trying to unstick it.
There is also a quiet lesson in how this round came together. Strategic capital showed up. Synopsys and Cadence are not tourists. Alchip is not guessing. SoftBank does not wander. When that mix aligns, it usually means the problem is real and the solution is close enough to touch.
