Earth Compute
general purpose As of August 25, 2026

Server CPUs

328.37 exaFLOP/s
0.4% of the world's compute · accreted

Xeon, EPYC, Graviton and the rest — the silicon that actually runs the internet. Tiny in FLOP/s next to a GPU rack and indispensable anyway: databases, networking and orchestration are latency problems, not throughput problems.

Installed compute
328.37 EFLOP/s
peak dense BF16 FLOP/s
Change, 1 year
+13.3%
331.9k H100e today
Share of world
0.4%
of total installed compute
In the headline
Yes
~35.0% utilized
← All constituents

Server CPUs over time

Installed compute in zettaFLOP/s, log scale, 2008 → present. A straight line on this axis means steady exponential growth; the slope is the doubling rate. History before the 2026 anchor is reconstructed — see the methodology.

How it's counted

~120M active servers x ~2.5 TFLOP/s dense FP16 (AVX-512 / AMX / SVE across sockets).

~120 million servers. Intel AMX and Arm SVE2 give modern sockets real matrix throughput; most of the installed base predates them.

Overlap treatment: Host CPUs in accelerated nodes are counted here; their attached GPUs are counted in the AI lines. The two are separate silicon.

Growth: compounding at 1.15× a year (59.5-month doubling) between published vintages.

Source IDC server tracker; vendor datasheets

Full methodology →