There's a reason Silicon Valley VCs prefer software startups: they're way, way cheaper and more agile than hardware.

As the story says, it took Oxide two years to build their networking switch, and that's just one part of their system!

Despite Silicon Valley's success starting from hardware (... it's in the name), since at least the dot-com boom the money focuses on software. See also https://a16z.com/why-software-is-eating-the-world/

There's still money for hardware (in case the VCs hit the jackpot on the next Cisco or Apple), but it's much harder to get their attention than for a web (... or AI) startup.

So really, kudos to the Oxide folks for pulling off shipping a product! I'm rooting for you.

It's also easier to evaluate the execution competency of a hardware startup team

They did a good job of describing what Oxide's product-market fit was at the beginning of this article, which I always wondered about since I couldn't see how Oxide could fit in.

> The Oxide rack comes fully assembled; it just needs to be slotted in at a data center.

Does it slot into a standard data center rack, like a really tall old blade server?

Or does it need floor space and hookups separate from the aisles of standard racks?

It is its own separate rack.

I guess "PAID" at the top means you have to pay to read this post, not that Oxide paid them to write it.

That is correct.

The silkscreen of the first prototype service processor they show is the most fun.

“the switch is often a nexus of reliability and performance issues”

I see what you did there..