3 ms·
Why do you say it's technically unfeasible? The articles I saw "debunking" Bloomberg are pretty weak on their arguments. Even if Bloomberg got some things wrong
by smolder 6y ago
Why do you say it's technically unfeasible? The articles I saw "debunking" Bloomberg are pretty weak on their arguments. Even if Bloomberg got some things wrong, a tiny chip that subverts motherboard firmware to phone home for instructions is possible in something rice sized, for a sophisticated state actor. Network stacks already exist in firmware. You'd need to inject maybe 4KB of instructions somewhere, but I can't imagine it'd be many orders of magnitude more than that to bootstrap such a process.
The suspicious silent non-retraction raises a red flag for me, that there may have been some kind of intervention to stop their reporting on the topic.
- whoisjuan 6y agohttps://www.servethehome.com/investigating-implausible-bloomberg-supermicro-stories/ https://www.servethehome.com/investigating-implausible-bloom... This is a good explanation. This article is weird because it's paginated. Make sure to use the paginator below the body to read it fully. There are many good arguments in that article but the simplest one is that there's simply no modern lithography methods to produce a chip of that size with processing power, signaling interfaces and networking, which are all required to create a trojan ship like the one described by Bloomberg. Additionally you need to provide power to this hypothetical nano-chip so you need more components like capacitors which of course is never mentioned in the Bloomberg article.
- smolder 6y agoThe chip doesn't need networking, it needs to infect existing firmware with networking. The motherboard/nic firmwares already have that. Nor would it necessarily be limited to the IPMI side nic as they imply. Steganography could be used to get through firewalls on net facing NICs. The part about "most sensitive code" wrongly assumes that means on storage devices and not motherboard firmware. As for the size, there are no commercial lithography methods with that capability, perhaps. It'd be possible to put significant capacitance and logic in a very tiny package with advanced lithographic techniques, say like gates printed into layers of a small cap? Yeah I'm getting really out there, but to say there's nothing so interesting in the highly compartmentalized world of spy tech is to make a claim in ignorance, naturally. Unless you're in a position to say precisely what technology they have, in which case, please contact me privately, since I'm curious. ;)