Hacker News Reader

Comment by BrandiATMuhkuh | Hacker News Reader

What surprises me with Infineon is that they seem to not work on a LLM on a chip. In particular that here in Linz Infineon, Apple's 5G modem designer and the Johannes Kepler University with Sepp Hochreiter (Inventor of LSTM) are just next to each other. I watched a interview of Infineon head of AI just recently, and I was like excited to hear about some cool hardware. And all he did is talk about how he uses LLMs.

View on HN

AFAIK, they do power and analog stuff and digital logic for automotive and industry, not cutting edge digital logic at the smallest feature sizes.

Yeah sadly. I think they have all the know how and people and connections to pull that off. Yet as so often, Austrian (Linz) companies are all about being a supplier not a disrupter.

> Yet as so often, Austrian (Linz) companies are all about being a supplier not a disrupter.

That makes a lot of sense when you consider the origins. After all, Infineon used to be a subsidiary of Siemens, a famously relatively conservative company

Infineon has these weird automotive tricore microprocessors. Going into consumer market it would be expensive and not competitive. Infineon tried that with XMC microprocessors, but they weren’t cheap, had thermal issues and weren’t well received by customers.

Does somebody work on combining flash storage technology with analog computers? After all, a flash storage cell is "just" a FET, so in theory it should be possible to use the cell directly as a weight of an input to some summing amplifier. In other words combine some kind of opamp with a bunch of flash cells to make up one artificial neuron. Should massively increase density and energy efficiency compared to traditional digital stuff. And slight inaccuracies of analog circuits won't matter that much in artifcial neural networks. AFAIU.

Answering my own question, googling yields a bunch of links to different approaches doing that and the technological challenges that are involved.

Someone needs to still keep making the "literally everything else" chips in the world. I'd rather it be an old name like Infineon instead of them handing their entire business to China in favor of chasing a bubble.

What's interesting to me in the last ten years we've seen a shift to silicon carbide and gallium nitride power components.

Those are key components in the shift to renewable power. Inverters for solar, motors, battery chargers, even inverter based power transformers.

Why would they? Their bread and butter is power electronics. Those big GPU clusters need dozens if not hundreds of heavy duty GaN/SiC FETs.

Going into something so far out of their expertise would be (rightfully?) punished by the market. That kind of innovation is more likely to succeed in a startup than in a traditional lifestyle company like Infineon.

Their black magic is a different breed.

Reply on news.ycombinator.com