AI Term of the Moment

chatbot


Look Up Another Term


Definition: Project Colossus


SpaceX data centers for AI training and running the company's Grok models. The first Colossus data center began operating in 2024 in a revitalized Electrolux vacuum cleaner factory in Memphis, Tennessee, where it trained xAI's Grok-3 chatbot.

Colossus 2 in Southhaven, Mississippi, about 15 miles south of Memphis, is partially operational as of 2027. A third datacenter will be built in Southhaven, and collectively all three are expected to house more than a million NVIDIA GPUs. See xAI and AI data center.

Colossus and Terafab
Colossus is the xAI data center and Terafab is Musk's semiconductor manufacturing facility (see Terafab). See Project Stargate.






100,000 Liquid Cooled GPU Modules
The first Colossus data center went into operation with 100,000 NVIDIA H100 modules (top). Colossus is expected to double in size to 200,000 modules. SuperMicro's water cooling keeps the required temperature. See H100 and Blackwell. (Image courtesy of xAI, Supermicro Computer, Inc. and NVIDIA.)







A Massive Network
Networking all this equipment is no simple feat. An RDMA fabric ties all servers together in one huge Ethernet running at 400 Gbps (see RDMA and Terabit Ethernet). (Images courtesy of xAI and Supermicro Computer, Inc.)






Launched in Only Four Months
Phase I required 150 MW (megawatts) of electricity. Combined with 8 MW from the local grid, 14 transportable VoltaGrid generators were installed to launch the data center in July 2024 with a third of the H100s. To keep the power uniform, all electricity flows into batteries and then to the equipment. (Image courtesy of VoltaGrid LLC.)





A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
X
Y
Z
misc