No GPU dependency
Jam runs on existing CPU + NVMe. Pre-Jam compression hit a ceiling around 2–3×; we hit 3–8× on the same iron. No new accelerator, no waiting on supply.
PRODUCTS
A sovereign cloud, compressed at the data plane. Indexed by a Knowledge Graph.
Cithorum sells one core product — a managed 1 PB Indian data centre powered by Jam (our compression / storage codec) and made queryable by the Data-Centre Ops Knowledge Graph. The same Jam codec is also licensed to other operators. The same KG core is the seed for vertical SaaS extensions (Workspace, Law, Medical-Sequencing) when paid demand justifies them — but the live product today is the data centre.
01 — THE THESIS
Storage gets cheaper every year. Data grows faster.
Storage hardware drops 25–30% in cost per year. Enterprise data grows 60–80% in the same window. The gap widens forever. Anyone running the numbers ends up paying more in real terms each cycle to hold what they already own. The hyperscalers turn that gap into margin. Everyone else turns it into burn.
This is a software problem, not a hardware one. Compression at the codec level — zstd, gzip, 7zip — hit a ceiling years ago. Codecs squeeze bytes inside a single file; they do not understand layouts, deduplicate across files, or index for restore. Jam is the next layer. It compresses end-to-end on existing CPU + NVMe, with no GPU dependency, and lands 3–8× on real production workloads on top of whatever codec the customer already runs.
Cithorum wins because the moat is operational, not algorithmic. The SDK drops in alongside existing storage — no rebuild, no data migration, customer-controlled deployment. Throughput is NVMe-bound, not algorithm-bound. We have real production proof: M2M TechConnect live at $7k/month MRR on the Starter tier, and a 1 PB proof environment in build that the next round of customers walk into.
THE PRODUCT
One core. Two compounding layers.
Cithorum Cloud is the live product — a managed 1 PB sovereign cloud, powered by Jam at the data plane and indexed by the Data-Centre Ops Knowledge Graph. Jam is also licensed standalone to other operators. The KG core extends into vertical SaaS later. Same substrate throughout.
01 · CITHORUM CLOUD · LIVE
The 1 PB Indian data centre. Jam at the data plane, Ops KG on top.
See Cithorum Cloud →02 · JAM ENGINE · LIVE
The compression / storage codec. Powers Cithorum Cloud and licensed standalone.
Learn about Jam Engine →03 · KNOWLEDGE GRAPH · OPS LIVE
Data-Centre Ops KG today. Vertical SaaS extensions next.
Explore the KG →Where the KG extends next — Workspace KG (mid-market revenue ops), Law KG (matter memory + audit-trailed AI), Medical-Sequencing KG (genomics labs and clinical research). Each is a different schema and a different buyer, but the core engine is the same one running Cithorum's own data centre today. They will be developed sequentially, design-partner-led, one at a time.
02 — WHY WE WIN
Software at the layer the codecs gave up on.
Drop-in SDK
S3-compatible gateway, Linux daemon, no rebuild. Customer keeps existing storage and applications. Two-to-four weeks from kickoff to a benchmark report on the customer's own data.
Customer-controlled
Data never leaves the customer's perimeter unless they choose. Restore is deterministic and verified — every byte hash-checked on the way back. Air-gapped option for sovereign deployments.
Production-proven
M2M TechConnect at $7k/month MRR. NVIDIA Inception programme. NATO BRAVE1 vetting. RTX/Raytheon ecosystem approval. ZenLaunchpad CAD $30k won. Atreides IO partnership in flight.
03 — HOW WE SELL
From qualify to signed MRR in eight weeks.
Six steps. Each step has a week, an owner, a deliverable. The same flow runs every Jam pilot and every Data Centre Service engagement.
01
Qualify workload
Fit assessment + dataset characterisation.
02
Sign pilot terms
MSA + benchmark-rights frame.
03
Provision access
Bucket, credentials, quota.
04
Run Jam
Ingest, compress, restore, meter.
05
Report proof
Benchmark report (anonymised public + private detailed).
06
Convert to paid
Production contract, MRR live.
04 — WHY NOW
The window opens once. We are inside it.
Data growth is non-linear
Compute cost is dropping; storage is the bottleneck. Codecs hit their ceiling years ago. Software is the next leg, and Jam is the only software-only entrant in the bracket with a live paying customer.
Hyperscalers are vulnerable
Cold-tier pricing hides egress and restore costs. Customer-controlled infrastructure is in fashion — sovereignty, regulation, cost. Buyers are looking for the exit; we sell the alternative on their own iron.
AI workloads make this worse
Training datasets are doubling; inference data is exploding. Whoever owns the data plane underneath wins compounding margin. Jam is that data plane — compressed, indexed, and accelerated for AI loading.