Brit startup Graphcore tossed a £200m early Christmas pressie for machine learning CPU AI chip startup Graphcore has managed to grab $200m from investors, part of which will be used to hire more engineers on the west coast...the west coast of Britain for the absence of doubt. Suffering ceepie-geepies! Do we need a new processor architecture? READ MORE The Bristol-based biz develops Intelligence Processor Unit ( …
You know if you Need a general anaesthetic, they put you out with a needle, this doesn't cause you to forget everything, you are "Parked". Same would go with this kind oif technology, turning off would "Park" it, but its state would be preserved by solid storage. Both humans and machines ready to be unparked, ready to fight another day (if all goods well with the I operation!). So yes it is ethical, because you aren't destroying it.
So, despite not having sold any material volume of anything, it is worth the fat end of two billion bucks?
A word of advice to the founders: Cash out now. VCs and their cash are easily parted, but eventually they'll smell a theranos that they've trodden in.
Engineers will be hired for the Bristol HQ and offices in London, Oslo and Palo Alto. New offices will be opened in Beijing, China, and Hsinchu, Taiwan. Graphcore said it will also expand its production capability and sales, marketing, finance, legal and HR headcount.
Not to mention a fleet of Tesla company cars, pool tables, free fruit and coffee, naked office assistants, at-desk massage, a company yacht, and a fancy new headquarters building in the most fashionable quarter of Bristol.
Hold on! Haven't I seen this move before? You know, that one where investors persuaded themselves that somebody had invented a machine that defied the laws of physics?
"Hold on! Haven't I seen this move before? You know, that one where investors persuaded themselves that somebody had invented a machine that defied the laws of physics?"
Heck, they only promise a 10x boost over x86, for the specific targetted workloads they will be used for. These are not general purpose CPU's, so being 10x faster in machine learning applications is probably a bit low on what could be achieved. I'd rank this one as a fairly safe bet...
"A word of advice to the founders: Cash out now. VCs and their cash are easily parted, but eventually they'll smell a theranos that they've trodden in."
They've only been going for two years, they should be able to rinse the VCs for at least 4-5 years before they realise that the company has no revenue whatsoever. Thernaos made it from 2003 all the way up until 2018. Uber is still going despite losing money hand over fist every year.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
