COO at Digital Storm | High-Performance Workstations, AI Compute & PC Hardware Infrastructure
Bay Area, CA
Joined December 2014
- Tweets487
- Following49
- Followers184
- Likes1.3K
Check out this clean build we did for a customer, featuring an @AMD Ryzen 9 9950X3D, 64 GB of @kingstontech 6400 MT/s, and an @NVIDIAGeForce RTX 5090 water cooled.
One of the biggest winners of the AI boom started out making kitchen drawer slides.
Taiwan's King Slide now reportedly controls 70–80% of the high-end AI server rail market, with an 87.4% gross margin in Q2, which is higher than NVIDIA.
Wild how a <$100 mechanical component can become so critical to AI infrastructure.
4
187
Not every build needs an RTX Pro 6000! This customer chose to go with the RTX Pro 4500 and got three of the same build for their in-house team.
3
22
2,210
Are gaming GPUs becoming an afterthought?
NVIDIA GPU prices are reportedly rising again, while Intel is rumored to be shifting focus from its Arc B580 toward higher-margin AI products.
When AI hardware is far more profitable, how long will gamers remain a priority?
1
3
195
Looking back at my old pics and came across these Titan Z.
I miss when NVIDIA made stuff like this.
1
115
What caught my attention in robotics this week was a company offering the intelligence separately from the robot.
@dreamlabs_ai launched an API that takes camera input and instructions and returns robot actions. It says it already works with more than 20 hardware partners. The current applications are focused on tabletop tasks, so there’s still plenty to prove before treating this as factory-ready automation.
Currnetly, PC manufacturers don’t have to develop their own processors, operating systems, and applications to build a product. We select the components, integrate them, validate the system, and support it for the customer’s workload.
I’m curious whether more of robotics could develop that way. One company builds the arms and grippers. Another supplies the model. An integrator makes the combination useful for a specific operation.
On a production floor, that last part matters. You still need to handle the actual packaging, placement, inspection requirements, and exceptions involved in the job. Access to a capable model could lower one barrier while leaving a substantial integration business around it.
It also creates a purchasing question: how easily can the customer improve the robot’s capabilities without replacing the machine?
83
Just got an email saying that DDR5 memory is increasing 10-15% this week and previous placed POs that haven't delivered yet are cancelled.
Just for record: we are honoring our customers for any previously placed orders at old pricing, but this market is getting even crazier!
4
89
There's a lot of expensive GPU hardware sitting idle between workloads. If providers can make that capacity easier to access without sacrificing reliability, it could open up a much larger market than traditional GPU rentals alone.
1
3
257
