Conventional AI
Software simulates neural networks on general-purpose processors. The network is a program running on hardware built for arithmetic.
Building the hardware of intelligence.
SYNTHETIC BRAINS
WE DON'T SIMULATE THE BRAIN. WE BUILD IT.
Synthetic neural systems designed to learn, adapt and compute in hardware — engineered from physical artificial neurons and synthetic synapses.
An early-stage deep-tech research company building a new substrate for intelligence.
In motion
A first look at the substrate
A new substrate for intelligence
Today's AI runs on hardware built to calculate. We are building hardware designed to think.
For decades, neural networks have been simulated in software, running on processors that were never designed for them. Every thought is emulated, one instruction at a time, at enormous cost in energy and speed.
Synthetic Brains takes a different path. Instead of simulating neurons in software, we build them in silicon. The neural architecture itself becomes physical.
“What if the network wasn't a program — but the machine itself?”
- Artificial neuron
- Synthetic synapse
- Neural substrate
- Adaptive network
- Synthetic brain The research frontier
Software simulation
- Weights stored in memory
- Fetched to the processor
- Multiplied step by step
- Repeat billions of times
Physical neural hardware
- Neurons are devices
- Synapses hold their own state
- Signals flow through the material
- Memory and compute co-located
The key distinction
We are not growing brains. We are building them.
Three approaches, three very different machines. Synthetic Brains occupies the space no one else does — where the neural architecture is physically engineered.
Biological computing
Living neurons are cultured and wired to electronics. Powerful, but constrained by biology, viability and control.
Synthetic Brains
The neural architecture itself is manufactured — artificial neurons and synthetic synapses fabricated as physical devices. No simulation. No living tissue.
The equation
- Artificial neurons
- Synthetic synapses
- Physical neural architecture
- Synthetic brain
This is an advanced research platform and computing architecture. We make no claim of human-level intelligence, consciousness, or a functioning artificial human brain.
From component to system
How a synthetic brain is built
Complexity emerges from the physical architecture itself. Simple engineered components combine, layer by layer, into systems where computation, memory and adaptation live in the hardware.
- Artificial neuronA physical element that reproduces key properties of a biological neuron
- Synthetic synapseAn adaptive connection whose state changes with activity
- Neural substrateNeurons and synapses interconnected across a semiconductor fabric
- Adaptive networkSignals propagate; connection strengths reorganize with use
- In-memory computationProcessing and memory occupy the same physical structure
- Emergent behaviorLearning and state arise from the architecture, not from code
- Synthetic brainA physical system that learns, adapts and computes in hardware
The technology
Three physical building blocks
Everything is engineered, measurable and manufacturable. Under magnification the structures echo a nervous system — but every element is a fabricated device.
Artificial neuron
A physical electronic element engineered to reproduce important properties of a biological neuron — integrating inputs and firing when a threshold is crossed.
Synthetic synapse
An adaptive connection whose physical state changes with activity. It stores its own weight — memory and computation in a single device.
Neural substrate
Neurons and synapses interconnected across a semiconductor fabric. Computation, memory and adaptation emerge from the physical architecture itself.
Imagery and diagrams are illustrative representations of engineered structures, shown for concept communication.
The reference design
Biology already engineered the most efficient neural hardware
A biological nervous system is not software. It is physical architecture — where memory, computation and adaptation are built into the same material. That is the design principle we are translating into engineered devices.
- Massively parallel
- Sparse, event-driven
- In-memory computation
- State held locally
- Adaptive connections
- Extreme energy efficiency
“Biology solved the architecture. Our task is to build it from materials we can manufacture.”
The convergence
Where different fields become one
Synthetic Brains sits at the intersection of multiple rapidly evolving disciplines.
Synthetic
Brains
Brains
- Neuroscience
- Microelectronics
- Material science
- Neuromorphic computing
- Device physics
The long-term vision
A new substrate, and the machines it could power
“When intelligence lives in the hardware, it can go everywhere hardware can go.”
The long-term vision is not to reproduce a human brain device by device.
The goal is to build a physical neural substrate — efficient enough, adaptive enough and small enough to bring learning directly into the machines around us.
Potential long-term applications
- Always-on edge intelligence
- Autonomous robotics
- Low-power sensing
- Adaptive medical devices
- Scientific instruments
- On-device learning
- Space & extreme environments
- Industrial control
- Next-generation computing
A possible trajectory — illustrative, not a promise
Devices
- Artificial neuron elements
- Synthetic synapse devices
- Materials characterization
- Single-device measurement
- Fabrication methods
Substrates
- Interconnected arrays
- Analog in-memory compute
- On-substrate adaptation
- Yield and reliability
- Hardware/software co-design
Systems
- Scaled neural substrates
- Learning in the hardware
- Energy-efficient inference
- Application-specific synthetic brains
The long horizon
“Intelligence becomes an engineered material.”
Illustrative research horizon, highly uncertain. These dates are not predictions or promises.
The people
A team built at the intersection
Neuroscience, device physics and semiconductor engineering: the three disciplines synthetic neural hardware demands, under one roof in Tel Aviv.
- Dr. Maya AronovCo-founder & CEO PhD, Neuroscience, Hebrew University of Jerusalem; neural signaling & device design
- Dr. Eyal ShenhavCo-founder & Chief Scientist PhD, Electrical Engineering, Technion; memristive devices & neuromorphic circuits
- Nir CohenFounding Engineer Ex-NVIDIA hardware engineer; semiconductor test & hardware/software co-design
For investors & partners
Build the infrastructure of the next intelligence era
The largest technological shifts rarely begin with a finished product. They begin with a question that eventually becomes an industry.
- What if the neural network were the chip, not a program on it?
- What if memory and computation shared the same device?
- What if learning happened in the material itself?
- What if intelligence could be manufactured, not just trained?
Synthetic Brains is building synthetic neural hardware at the intersection of neuroscience, microelectronics and material science — a new substrate for intelligent machines.
Why this could matter
Potential sources of long-term differentiation
- Proprietary device physics
- Novel materials
- Synapse architectures
- Fabrication processes
- In-memory compute designs
- Characterization data
- Hardware / software co-design
- Foundry partnerships
- Patent portfolio
- Deep interdisciplinary talent
These represent potential future defensibility, not claims of an existing competitive moat.
Research principles
Evidence before hype
- Evidence before hypeClaims follow results, not the other way around.
- ReproducibilityFindings should hold when others run them.
- Measurable device metricsProgress defined by what we can measure on the bench.
- Energy efficiencyEfficiency treated as a first-class objective.
- In-hardware adaptationState and learning that live in the physical substrate.
- CharacterizationUnderstanding why a device behaves as it does.
- Safety by designSafety considered from the architecture up.
- Human oversightPeople remain in the loop and in control.
Synthetic Brains is a forward-looking research concept. The technologies described represent areas of investigation and long-term research direction, not claims of completed scientific achievement.
Questions worth building for
Questions worth building for
What if a neuron could be a physical device, not a line of code?
What if a synapse could hold its own memory?
What if computation and memory lived in the same material?
What if a network could learn without being reprogrammed?
What if intelligence could be fabricated, wafer by wafer?
What if the next leap in AI isn't a larger model?
What if it is a new kind of machine?
A human purpose
Intelligence should expand human possibility
The objective is not to replace human intelligence. It is to build a new substrate for machine intelligence — efficient and adaptive enough to expand what humanity can discover, create and solve.
- Discover faster
- Create more
- Understand deeper
- Solve harder problems
- Explore further
The hardware of intelligence
has not been built yet.
We think it starts on a lab bench, not in a data center. Come build it with us.
Message received.
Thanks, there. We'll be in touch soon.