# Scaling Intelligence: Designing Data Centers for Next-Gen Language Models

arXiv: https://arxiv.org/abs/2506.15006 · HTML: https://mlsanity.com/paper/2506.15006/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 5 of its 34 references that are in the map.
- First posted on arXiv 17 Jun 2025.

## Builds on (most cited first, 5 of 5)

- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints](https://mlsanity.com/paper/2305.13245/) (2023)
- [MegaScale: Scaling Large Language Model Training to More Than 10,000 GPUs](https://mlsanity.com/paper/2402.15627/) (2024)
- [Accelerating Distributed MoE Training and Inference with Lina](https://mlsanity.com/paper/2210.17223/) (2022)
- [Ridgeline: A 2D Roofline Model for Distributed Systems](https://mlsanity.com/paper/2209.01368/) (2022)

## Built on by (most cited first, 3 of 3)

- [Distributed Training under Packet Loss](https://mlsanity.com/paper/2507.07114/) (2025)
- [Designing Datacenter Power Delivery Hierarchies for the AI Era](https://mlsanity.com/paper/2605.16255/) (2026)
- [The Life of a Token: from Words to Bits on the Wire](https://mlsanity.com/paper/2609.19924/) (2026)

## Abstract (arXiv)

The explosive growth of Large Language Models (LLMs), such as GPT-4 with 1.8 trillion parameters, demands a fundamental rethinking of data center architecture to ensure scalability, efficiency, and cost-effectiveness. Our work provides a comprehensive co-design framework that jointly explores FLOPS, HBM bandwidth and capacity, multiple network topologies (two-tier vs. FullFlat optical), the size of the scale-out domain, and popular parallelism/optimization strategies used in LLMs. We introduce and evaluate FullFlat network architectures, which provide uniform high-bandwidth, low-latency connectivity between all nodes, and demonstrate their transformative impact on performance and scalability. Through detailed sensitivity analyses, we quantify the benefits of overlapping compute and communication, leveraging hardware-accelerated collectives, widening the scale-out domain, and increasing memory capacity. Our study spans both sparse (mixture of experts) and dense transformer-based LLMs, revealing how system design choices affect Model FLOPS Utilization (MFU = Model FLOPS per token * Observed tokens per second / Peak FLOPS of the hardware) and overall throughput. For the co-design study, we utilized an analytical performance modeling tool capable of predicting LLM runtime within 10% of real-world measurements. Our findings offer actionable insights and a practical roadmap for designing AI data centers that can efficiently support trillion-parameter models, reduce optimization complexity, and sustain the rapid evolution of AI capabilities.