Aug. 28 at 6:23 PM
$RGTI $QUBT $QBTS $IONQ $NVDA
This is what happens when an amateur tries to make intellectual interpretation (a pure guess pulled out of butt) without doing a properly deep research.
The high end GPU servers can last 5 to 10 years, when including downstack cascading older servers from LLM training clusters and reassign them to inference (running existing models).
The high end GPU servers can cost
$200K to
$250K, so that's cost of
$20K to
$25K a year. A ~95% utilized servers from hyperscalers and neoclouds can generate
$240K to
$740K revenue annually, and net profit of
$150K to
$250K.
You think all those high-paid, well educated executives at biggest companies in the world can't figure out the basic math and they decide to just gamble away their company.
Anyone following this fearmongering moron should unfollow and add him to garbage poster block list...like me.
