Pith. sign in

REVIEW 9 cited by

Trends in AI Supercomputers

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2504.16026 v2 pith:YE64KHJZ submitted 2025-04-22 cs.CY cs.AI

Trends in AI Supercomputers

classification cs.CY cs.AI
keywords supercomputerscosthardwareperformancepowertrendsneedssupercomputer
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Frontier AI development relies on powerful AI supercomputers, yet analysis of these systems is limited. We create a dataset of 500 AI supercomputers from 2019 to 2025 and analyze key trends in performance, power needs, hardware cost, ownership, and global distribution. We find that the computational performance of AI supercomputers has doubled every nine months, while hardware acquisition cost and power needs both doubled every year. The leading system in March 2025, xAI's Colossus, used 200,000 AI chips, had a hardware cost of \$7B, and required 300 MW of power, as much as 250,000 households. As AI supercomputers evolved from tools for science to industrial machines, companies rapidly expanded their share of total AI supercomputer performance, while the share of governments and academia diminished. Globally, the United States accounts for about 75% of total performance in our dataset, with China in second place at 15%. If the observed trends continue, the leading AI supercomputer in 2030 will achieve $2\times10^{22}$ 16-bit FLOP/s, use two million AI chips, have a hardware cost of \$200 billion, and require 9 GW of power. Our analysis provides visibility into the AI supercomputer landscape, allowing policymakers to assess key AI trends like resource needs, ownership, and national competitiveness.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. LLMSpace: Carbon Footprint Modeling for Large Language Model Inference on LEO Satellites

    cs.LG 2026-05 unverdicted novelty 7.0

    LLMSpace is the first modeling framework that jointly calculates operational and embodied carbon emissions for LLM inference on LEO satellites, incorporating radiation-hardened hardware, peripheral systems, and LLM wo...

  2. LLMSpace: Carbon Footprint Modeling for Large Language Model Inference on LEO Satellites

    cs.LG 2026-05 unverdicted novelty 7.0

    LLMSpace is the first framework to jointly model operational and embodied carbon for LLM inference on LEO satellites, incorporating radiation-hardened hardware, peripheral systems, and workload patterns such as prefil...

  3. On the Surprising Effectiveness of a Single Global Merging in Decentralized Learning

    cs.LG 2025-07 unverdicted novelty 7.0

    A single global merge at the final step of decentralized SGD matches the convergence rate of parallel SGD while improving test accuracy under high data heterogeneity.

  4. New Number Formats for FFT IP Cores in Optical OFDM Transceivers

    cs.AR 2026-07 conditional novelty 6.0

    Custom 11-12 bit floating-point FFT cores in a 128 Gbit/s optical OFDM transceiver match 16-bit fixed-point accuracy at lower power and area for moderate QAM orders, with degradation at 4096-QAM.

  5. StickyInvoc: Rethinking Task Models for High-throughput Workflows in the LLM Era

    cs.DC 2026-06 unverdicted novelty 6.0

    StickyInvoc introduces sticky tasks that load LLM model state once and invocation tasks that reuse it, yielding 3.6x speedup on a 150k-inference workflow.

  6. Communication-Semantic-Aware RDMA Loss Recovery for QP-scalable Hyperscale AI Training

    cs.NI 2026-05 unverdicted novelty 6.0

    CSA-UD is a communication-semantic-aware unreliable datagram RDMA loss recovery mechanism that improves QP scalability and reduces 99th percentile flow completion times in hyperscale AI training collectives.

  7. Switching Efficiency: A Novel Framework for Dissecting AI Data Center Network Efficiency

    cs.NI 2026-04 unverdicted novelty 6.0

    Introduces Switching Efficiency (η) decomposed into data, routing efficiency, and port utilization factors to analyze and improve communication bottlenecks in AI data center networks for LLM training.

  8. How Sovereign Is Sovereign Compute? A Review of 775 Non-U.S. Data Centers

    cs.CY 2025-07 unverdicted novelty 6.0

    U.S. operators control 48% of non-U.S. data center projects by investment value, limiting digital sovereignty for host nations and offering the U.S. an additional governance tool for deployed AI infrastructure.

  9. Extreme-Scale Interconnection Networks

    cs.NI 2026-05 unverdicted novelty 4.0

    MRLS leaf-spine networks deliver 50% higher throughput than Fat-Tree and 100% higher than Dragonfly for All2All collectives with 100k endpoints via simulation evaluation.