Expanse

Roadmap & Position in GPU Infra

Finds wasted GPU and HPC capacity before jobs burn money.

Company Overview

Expanse is a compute intelligence layer that predicts GPU and HPC job needs, cuts cluster waste, and diagnoses failed runs. The buyers are platform teams in AI infrastructure, quant finance, drug discovery, and research labs; named customers are not public.

What They're Building

The company's public product roadmap & what they're committed to building.

Passive Telemetry

Cluster agents capture job requests, runtime metrics, logs, and outcomes so resource advice is based on actual workloads, not assumptions.

Resource Analysis

The analyse command predicts memory, runtime, GPU use, failure risk, and better job configs before a run starts.

Failure Diagnosis

The diagnose command turns failed jobs into root-cause notes with code and config changes a platform team can act on.

Waste Scanner

An open scanner checks SLURM and Kubernetes clusters for CPU, memory, and GPU waste, producing a shareable report.

Scheduler Integrations

Expanse supports SLURM, Kubernetes, Databricks, YARN, cloud batch systems, Nomad, Ray, Flyte, and related workload stacks.

Latest Intelligence

Zeitgeist tracks private signals to determine where the company is heading strategically.

Cluster Waste Becomes Prediction Layer

May 14, 2026

Confidence:

Medium

New Intel: Expanse is turning cluster telemetry into pre-run resource prediction and failure diagnosis. That opens the GPU waste budget and pressures Run:ai, CAST AI, and homegrown platform tooling.

Founder and Key Execs

Ismaeel Bashir

Co-Founder & CEO (training workloads at quant funds and national supercomputing centers, EPCC resource prediction work)

Nikodem Bieniek

Co-Founder & CTO (hedge fund infrastructure and GPU cluster speech ML experience)

Yafet Melake

Co-Founder & COO (quant infrastructure tooling and graph neural network work for SLURM queue prediction)

Eren Mendi

Co-Founder & CPO (quant finance prototyping and ML systems work on decentralized foundation model training)

Founder Force Multiplier

The four-person founding team combines HPC scars (Ismaeel Bashir at quant funds and national supercomputing centers), quant infrastructure taste (Nikodem Bieniek at hedge funds), graph neural network work on SLURM queue prediction (Yafet Melake), and ML systems experience on decentralized foundation model training (Eren Mendi). That combination should land with buyers who distrust generic AI wrappers and care about scheduler reality.

Funding History

2025 | Founded
2026 | Joined Y Combinator Spring 2026 batch

Competitors

NVIDIA Run:ai:

GPU orchestration platform with NVIDIA distribution; competes on cluster utilization, but is closer to scheduler control than neutral workload intelligence.

CAST AI:

Cloud and Kubernetes cost optimizer; stronger in cloud spend automation, less focused on HPC job prediction and failure diagnosis.

HPE Determined AI:

ML training platform with resource management; broader experiment workflow, weaker fit as a scheduler-agnostic intelligence layer.</p>

Expanse

's Moat:

The credible path runs through proprietary workload telemetry combined with workflow switching costs that emerge once predictions are trusted inside SLURM and Kubernetes queues. Neither is proven yet.

How They're Leveraging AI

Resource Prediction

Predicts GPU and HPC job resource needs before a run starts, including memory, runtime, GPU use, and failure risk.

Failure Diagnosis

Diagnoses failed cluster jobs and suggests config or code changes tied to the evidence from logs and telemetry.

Feedback Loop

Builds a customer-specific workload memory so recommendations improve as more jobs run through the cluster.

AI Use Overview:

Expanse uses passive cluster telemetry, runtime metrics, and historical job outcomes to train resource-prediction models that operate inside SLURM and Kubernetes queues, which is a more grounded approach than generic AI cluster wrappers.

More Similar Companies

Entire

Git-native AI code explainability and session context capture

The ex-GitHub CEO is building the compliance layer for AI-generated code, with personal relationships to every enterprise buyer who will need it.

Pinecone

Managed vector database and knowledge infrastructure for production AI apps.

A category winner pitch rests on Pinecone turning vector search into the default memory layer for RAG, agents, and enterprise knowledge apps.

Approxima

Lets product teams go from idea to deployed software in under an hour with AI agents.

Most AI coding tools target greenfield features. Approxima goes after the unglamorous maintenance work (bug fixes, incremental updates) that eats 60%+ of engineering time, with sandbox validation that lets agents merge to production without human review.

21st Labs

Helps developers ship AI apps 10x faster with purpose-built components and agent tools.

AI coding tools need a trusted component layer to ship production-ready UI, and their 1.4M developer distribution gives them a head start before Vercel or GitHub bundle one in.

Back To All Companies >