Network-on-Chip Fabric IP  ·  Bengaluru, India

The on-chip fabric that thinks at runtime.

Srotas is AI-native Network-on-Chip IP for inference silicon. It routes with workload awareness, accelerates collective communication in the fabric itself, and ships with formal proofs of correctness — not just testbenches.

Fabless IPlicensing model
AI inferenceMoE & multi-tenant serving
Chiplet-readyUCIe multi-die extension
Formally verifiedproofs shipped with the IP
The one-line thesis

Others automate the design. Others optimize the plan. Srotas makes the fabric intelligent at runtime — and proves it.

Why now

AI inference broke the interconnect.

Modern accelerators are no longer one chip running one model. They are chiplet systems serving many tenants and Mixture-of-Experts workloads at once — where traffic patterns shift every few microseconds. A fabric that was configured once, at design time, can't keep up. The interconnect, not the compute, is now the bottleneck.

Workload
MoE serving creates bursty, all-to-all expert-routing traffic that defeats statically-tuned fabrics.
Tenancy
Multi-tenant inference demands hard isolation and per-tenant quality of service inside the chip.
Scale
Chiplet disaggregation stretches the fabric across dies — latency and ordering guarantees must survive the boundary.
Trust
Licensees bet the tapeout on third-party IP. Simulation coverage alone is no longer a good enough answer.
The Srotas fabric

Three things no other NoC IP does.

Srotas is built around three structural differentiators — each one architectural, not incremental, and each one aimed squarely at AI inference silicon.

01 · Runtime intelligence

The compiler tells the fabric what matters.

Compiler-emitted slack tags travel with the traffic, so the fabric knows which packets are latency-critical and which can wait. NIU-resident logic acts on that knowledge every cycle — no host in the loop, no design-time guesswork.

02 · Collective-native topology

Broadcast is a wire, not a workaround.

A concentrated mesh with MECS express channels makes one-to-many and reduction traffic first-class citizens. In-network reduction units collapse collective operations inside the fabric instead of bouncing them through memory.

03 · Proof as product

We ship theorems with the RTL.

Deadlock freedom, ordering, isolation — delivered as formal correctness proofs alongside the IP, born from a decade of design-verification practice. Your sign-off review starts from proofs, not from hope.

Architecture at a glance

A fabric designed for how AI silicon actually behaves.

Srotas is a scalable mesh-based fabric with express channels for one-to-many traffic, workload-aware scheduling at the network interface, hardware-enforced isolation between tenants, and in-fabric acceleration of collective operations. It extends cleanly across chiplet boundaries and is developed with formal verification as a first-class deliverable. Detailed architecture documentation is available to licensees and partners under NDA.

Topology
Concentrated mesh with express channels tuned for AI traffic patterns
Intelligence
Workload-aware runtime scheduling driven by compiler hints
Isolation
Hardware multi-tenant partitioning with per-tenant guarantees
Multi-die
Chiplet-ready — fabric semantics preserved across die boundaries via UCIe
Verification
Formally verified — correctness proofs delivered alongside the IP
Research programme

Every claim runs through the pipeline first.

Architecture decisions at SrotasSemi are settled by data, not opinion. Our containerized T3 toolchain evaluates every topology and routing choice end-to-end — from workload mapping to Pareto frontier.

T3 / STAGE 1

Timeloop

Maps real AI workloads to the accelerator and extracts the traffic each layer actually generates.

T3 / STAGE 2

Traffic matrix

Converts workload behaviour into fabric-level traffic demands — the ground truth for evaluation.

T3 / STAGE 3

Booksim

Cycle-accurate network simulation of candidate topologies and routing policies under that traffic.

T3 / STAGE 4

Pareto dashboard

Latency, throughput, and cost trade-offs visualised — so the frontier picks the architecture.

Industry–academia collaboration. SrotasSemi runs a structured six-month NoC research programme with student researchers from NIT Karnataka, spanning workload modelling, network simulation, and design-space exploration — with results targeted at top-tier design-automation venues.
About

Interconnect people, building interconnect IP.

SrotasSemi is a fabless semiconductor IP company headquartered in Bengaluru, India — a city that houses a substantial share of the world's chip-design talent, yet almost none of the world's chip IP ownership. We exist to change that ratio.

The company was founded by a semiconductor engineer whose career spans Network-on-Chip architecture and design verification at Qualcomm, Tenstorrent, and Meta — the places where modern AI silicon interconnect was designed, broken, and fixed. That verification-first DNA is why Srotas ships proofs, not just RTL.

Model
Fabless IP licensing — soft IP, integration support, and verification collateral
Ecosystem
Aligned with India's Semicon 2.0 / DLI 2.0 design-incentive programmes
Academia
Active research programme with NIT Karnataka on topology optimisation
HQ
Bengaluru, Karnataka, India
Work with us

Licensing, partnership, or investment — let's talk fabric.

Whether you're architecting an AI accelerator, exploring chiplet interconnect, or backing India's semiconductor IP story — we'd like to hear from you.