01 — PrologueValsad, India
An independent semiconductor IP foundry

Adaptive intelligence,licensed into silicon.

NeuroCore gives GPUs, NPUs, TPUs and SoCs brain-like intelligence — event-driven, in-memory, and able to learn on-device. License it standalone, place it beside an accelerator, or embed it in your next SoC.

The operating principle

The leaders built extraordinary accelerators.
We make them think like a brain.

7nm – 45nm CMOSStandalone · co-processor · SoC IP
Scroll
02 — Thesis

Why we exist beside the leaders

GPUs,NPUsandTPUsareextraordinarymachines—themostsuccessfulacceleratorseverbuilt,andthereasonmodernartificialintelligenceexistsatall.NeuroCoredoesnotcompetewiththem.Itaddswhatdensematrixmathcannot:memorythatstayswherethearithmetichappens,eventsinsteadofaglobalclock,andlearningthatoccursinsidethesiliconin142microseconds,withoutatriptothecloud.

Complement

Adds to a GPU, NPU or TPU — never replaces it.

Continuity

State that persists across seconds, not frames.

Locality

Learning inside the tile in 142 µs.

03 — The leaders

The accelerators of this decade are extraordinary.

We hold them in genuine regard. They solved dense arithmetic at a scale that changed what computing is for. NeuroCore begins from that achievement — and adds the abilities a matrix engine was never designed to carry.

GPUGraphics processing unit

Parallel throughput without equal.

Massively parallel processors that turned neural networks from research into an industry. They remain the engine of large-scale training and the standard against which every accelerator is measured.

NPUNeural processing unit

Purpose-built efficiency.

Dedicated inference engines that put neural networks into phones, cameras and cars at remarkable performance per watt. They are the reason on-device intelligence became ordinary.

TPUTensor processing unit

Scale as an architectural idea.

Systolic arrays designed around tensor throughput, proving how far a single specialised workload can be pushed when silicon and compiler are designed together.

We build on their success — and add the qualities dense math cannot supply.

Temporal stateOn-device adaptationSparse event compute
04 — The memory wall
01The bus

Even the best accelerator pays a toll.

A GPU, NPU or TPU streams weights from off-chip memory to its arithmetic units, then streams activations back. The math is extraordinary. The journey is not.

Energy spent moving data
92%
Conventional
ACCELERATOROFF-CHIP HBMMILLIMETRES OF WIRE · EVERY WEIGHT
FUSED IN-MEMORY COMPUTE · ZERO OFF-CHIP BYTES
fig. 04
05 — NeuroCore
NeuroCore tile7nm – 45nm CMOS
01The tile

A tile, not a monolith.

One NeuroCore tile holds an array of processing elements, their private memory, and the routing between them. Repeat the tile and the array scales to the product you are building.

06 — Integration

Three ways to put a brain beside your accelerator.

NeuroCore is not a fixed product. It is licensed IP with configurable wrappers, memory maps, clocking and power domains — so the topology is entirely yours to choose, and to change.

01
DEVICENCSENSORACTUATOR

Standalone accelerator

NeuroCore as the primary inference engine in a dedicated device — no host GPU, NPU or TPU in the system at all. Sparse, event-driven workloads run at a power envelope a matrix engine cannot reach.

Best for always-on sensing, implants, instruments

02
SOC BOUNDARYCPUGPUNCAXI4 / TILELINK / OCP

Beside a GPU, NPU or TPU

Keep the leader for dense math. Add NeuroCore as a co-processor over AXI4, TileLink or OCP for temporal state, on-device adaptation and low-power inference between frames.

Best for automotive, defence, robotics

03
SOC BOUNDARYCPUISPNPUNCDSPMEM

Inside a new SoC

License NeuroCore as RTL and place it in your subsystem next to a CPU, ISP, DSP or an existing accelerator. Memory maps, clocking and power domains are configurable.

Best for custom silicon and heterogeneous SoCs

Flexibility is the product: bus wrappers, address maps, clock domains and power gating are part of the deliverable, so NeuroCore adapts to the ecosystem you already have.

See the interface specification
07 — Capabilities

What NeuroCore adds to an accelerator.

Four architectural decisions, each answering a failure mode of dense matrix engines.

01

Unified memory and MIMD parallelism

Processing elements sit inside their own storage. Event-driven threads execute asynchronously across the array, so parallelism comes from activity rather than a clock distribution network.

02

Local learning, in space and time

Weights adapt where the signal arrives. Local plasticity rules replace global gradient backpropagation, removing the activation cache that makes training expensive in conventional silicon.

03

Seconds of continuous context

Recurrent state lives natively in on-tile memory, so the core tracks continuous sensor streams and world models over seconds instead of isolated frames.

04

PyTorch and TensorFlow in, RTL out

Compiler tooling ingests standard models and maps them onto hardware-resident topology, so your design team keeps the flow and the models it already trusts.

Delivered as synthesizable RTL with C++ models, SDC constraints and verification collateral, ready for your ASIC or SoC flow.

Explore NeuroCore IP
08 — Evidence

The claims are in the silicon, not the slides.

Single-tile telemetry from evaluated RTL on FPGA and standard-cell flows — the same numbers our licensees receive in their evaluation package.

0 µs
On-the-fly adaptation
0×
Lower dynamic power
0.0%
Catastrophic forgetting
0.0 ms
Zero-shot transfer
At equal workloadConv.NeuroCore
Energy per inference1002
Off-chip memory traffic1000
Time to adapt on-device1001

Normalised to a conventional accelerator. Full methodology in the evidence dossier.

neurocore@tile-0 · serialverbatim

> run --benchmark=onchip_adaptation --tile=1

[INIT]synaptic array online · weights resident in SRAM · 0 memory cycles

[TRAIN]e-prop temporal gradient computed locally in space and time

[CONV]on-chip weight update finalised in 142.4 µs · peak die temp 34.2 °C

[EVAL]zero-shot inference 0.00 ms · off-chip bytes transferred 0

status: adapted & verified

09 — Frontiers
01
Automotive AI

Perception that adapts to the weather, not just the dataset.

Cabin monitoring and sensor fusion that keep learning on-device as cameras fog, lenses drift and light collapses. No uplink, no cloud re-validation cycle, no compromise on automotive safety process.

AIS-184 aligned30–50 °C passiveASIL-ready flows
02
Defence and aerospace

Cognitive RF inside a contested spectrum.

Jamming mitigation and radar de-noising resolved in microseconds on the platform itself, with the model adapting to the emitter it is actually facing — no ground uplink, no return to base.

Denied commsNo ground uplinkRadiation-aware PDKs
03
Industrial robotics

Reflex control on a micro-joule budget.

Sub-millisecond closed-loop actuation with on-chip learning that absorbs wear, payload change and tool drift, so a line keeps running instead of waiting for a recalibration stoppage.

Passive thermalWear-aware controlSub-ms loops
04
Edge vision

Intelligence where there is no thermal headroom.

Always-on event vision for cameras, drones and instruments: a spiking front end, resident weights, and a power envelope small enough to sit behind a battery for years.

Event-driven sensingBattery-scale powerAlways-on
Frontiers
04
10 — Licensing

You do not buy a chip. You license the core.

JSA LAABS is a fabless IP foundry. The licensee owns the silicon, the tapeout and the product. We deliver the architecture and stand behind it through closure.

01Deliverables

The IP core deliverable suite

Synthesizable Verilog and SystemVerilog RTL, cycle-accurate C++ models, SDC timing constraints and verification testbenches — everything a design team needs to evaluate the core inside its own flow.

RTL + C++ modelSDC / UPF collateralVerification testbenches
02Interconnect

Bus and subsystem fusion

Standard AXI4, TileLink and OCP wrappers connect NeuroCore to host processors, DMA engines and peripheral sensor interfaces, with memory maps, clock domains and power gating configured to your SoC.

AXI4 · TileLink · OCPHost CPU + DMA attachConfigurable memory map
03Silicon closure

RTL to GDSII enablement

Physical design guidance and timing closure support, validated across 7nm FinFET, 16nm, 28nm and 45nm standard CMOS process design kits, with signoff review alongside your team.

7nm – 45nm PDKsTiming closure supportSignoff review

Customers build the silicon. We deliver the architecture, the collateral and the engineering support to close it — for standalone parts, co-processor designs and heterogeneous SoCs alike.

11 — Questions

What licensees ask first.

Straight answers on integration, silicon and commercial terms.

01What is NeuroCore?

NeuroCore is adaptive AI accelerator IP from JSA LAABS. It is a synthesizable processor core that combines in-memory SRAM with event-driven, brain-inspired compute, so inference and on-chip learning happen where the weights already live.

02Can NeuroCore work with an existing GPU, NPU or TPU?

Yes. NeuroCore is designed as a complement to the industry's leading accelerators, not a replacement. It adds temporal state, on-device adaptation and low-power sparse compute beside a GPU, NPU or TPU over AXI4, TileLink or OCP interconnect.

03Can NeuroCore be licensed as a standalone processor?

Yes. It can serve as the primary inference engine in a dedicated device, or be embedded as one block among many inside a new SoC next to a CPU, ISP, DSP or an existing AI accelerator. The topology is the licensee's choice.

04Which process nodes and interfaces are supported?

The RTL is validated across 7nm FinFET, 16nm, 28nm and 45nm standard CMOS process design kits. Deliverables include AXI4, TileLink and OCP wrappers, SDC timing constraints, cycle-accurate C++ models and verification testbenches.

05How does licensing and integration work?

Engagements follow three stages: IP core deliverables, bus and subsystem fusion, then RTL-to-GDSII enablement. JSA LAABS supports timing closure and signoff review, while tapeout, fabrication and packaging remain with the licensee and its foundry partners.

06Where is JSA LAABS based?

JSA LAABS Pvt. Ltd. is an independent fabless semiconductor IP foundry based in Valsad, Gujarat, India, licensing processor IP to design teams, automotive programmes and defence primes worldwide.

EpilogueJSA LAABS · 2026

License adaptive intelligence into your next silicon platform.

Standalone, beside a GPU, NPU or TPU, or embedded inside a new SoC — the topology is yours. Start with an evaluation package or a technical briefing with the architecture team.

We build the IP · Our customers build the silicon

Valsad · Gujarat · India