Analog in-memory computing (AIC) chips developed by EnCharge AI (which integrates both computation and memory) provide an exponential reduction in the energy required for AI workloads from the edge to the cloud. EnCharge is targeting applications which have a need for high-performance computations in a power-constrained environment such as client devices, edge computing, and environmentally sustainable AI deployment. The company was founded by researchers at Princeton University and veteran semiconductor engineers, and its goal is to allow for advanced AI beyond what can be achieved using traditional cloud or GPU architectures.
EnCharge was formed by Princeton Professor of Electrical Engineering and Computer Science Naveen Verma and two former semiconductor engineers Echere Iroaga and Kailash Gopalakrishnan to commercially develop analog in-memory computing technologies researched at Princeton.
Initially raised $21.7M – $44M to develop core analog AI chip technology.
Received $100M in funding from Tiger Global, making total funding received equal to $144M for the commercialization of EnCharge’s first client AI accelerator.
In preparation for the market launch of EnCharge’s full stack AI solutions, EnCharge will enter the market with chiplets, ASICs and PCIe cards for use from edge-to-cloud.
Leverages existing semiconductor supply chain for diverse form factors
Analog in-memory computing architecture with fully validated silicon delivering breakthrough efficiency for Transformer-based and state-of-the-art AI models
Based on company website, technical announcements, and Princeton research background
| Service | Cost | Details | Source |
|---|---|---|---|
| Growth | $49/month | A/B Tests, Advanced Wait Steps, API, Built-in Templates, Calendly integration | GetApp, Capterra |
| Premium | $59/month | All Growth features plus Event-Based Segments, Salesforce Integration, Transactional Emails, Email Domains | GetApp |
| Business | $99/month for 2,000 subscribers | Advanced features for larger audiences, scales with user count | G2, ToolIndex, Official site |
| Premium Services Silver | $350/month | 4 hours of expert consulting included | encharge.io/premium-services |
| Premium Services Gold | $500/month | 8 hours of expert consulting included | encharge.io/premium-services |
| Premium Services Platinum | $900/month | 14 hours of expert consulting included | encharge.io/premium-services |
| Free Trial | Available | Free trial offered, no credit card details specified | SoftwareAdvice, G2 |
| Feature | Encharge | ActiveCampaign | MailerLite | Systeme.io |
|---|---|---|---|---|
| Core Functionality | Behavior-based email automation | Multi-channel marketing automation | Email marketing & newsletters | All-in-one funnel builder |
| Pricing (starting price) | $49/mo | No pricing info | No pricing info | No pricing info |
| Free Tier | No (free trial) | Yes | Yes | Yes |
| Enterprise Features | API, Salesforce integration | CRM, advanced segmentation | Limited | Basic |
| API Availability | Yes | Yes | Yes | Limited |
| Integration Count | Deep CRM + billing (Chargify) | 500+ | 50+ | Limited native |
| Support Options | Email, quick response | Phone, chat | Email, chat | |
| Security Certifications | — | GDPR | GDPR | GDPR |
Pricing transparency, minus the complexity. Encharge provides clearer pricing starting at $49/month yet lacks ActiveCampaign's scalability for CRM integration.
Encharge for SaaS product automations. ActiveCampaign for core marketing automation suites that need marketing stuff tacked on.
Advance segmentation, event triggers for product-led growth. Slightly pricier starting point ($49) than MailerLite’s free tier, but more powerful, particularly for automation.
Encharge for user flows. MailerLite for handling simple email campaigns.
Encharge, email automations. Systeme.io is a suite of sales funnels and everything sales. Encharge is for the team looking to automates their CRM. Systeme.io is cheaper for the solopreneur trying to put all things sales in a single system.
Encharge for marketing automation experts. Systeme.io for starting a cheeky boot-strapped company.
EN100 is EnCharge AI's first analog in-memory computing accelerator delivering 200+ TOPS for laptops (M.2) and 1 PetaOPS for workstations (PCI-E). It provides up to 20x better performance per watt than competing solutions. Available via Early Access Program for developers and OEMs
EnCharge uses charge-domain analog in-memory computing with metal capacitors, achieving 30 TOPS/mm² density vs 3 TOPS/mm² for digital architectures. This enables 20x energy efficiency for edge AI without cloud dependency. Proven across 5 generations of silicon
Laptops and workstations will be supported by M.2 version of the EN100 chip, which delivers over 200 TOPS at just 8.25W, and a PCIe version for workstation use with 4 NPUs reaching approximately 1 PetaOPS. The EN100 supports up to 128GB of LPDDR memory and can achieve 272 GB/s bandwidth. It's designed to power battery-operated devices and professional-grade AI applications.
The package includes a complete suite of AI development tools for optimizing code specifically for the EN100 and compiling popular frameworks like PyTorch and TensorFlow. It provides programming support for CNNs, transformer networks, encoder/decoder models, and more. The EN100 chip offers a fully programmable platform that works with today's AI models as well as those still in development.
Round One of the Early Access Program for the EN100 has closed to new registrations. If you're interested, Round Two registration is available at www.encharge.ai/en100 or env1.enchargeai.com. Early participants are already using the EN100 to build multimodal AI agents and gaming applications.
The EN100 enables on-device AI capabilities in laptops and workstations, including generative language models, real-time computer vision, always-on multimodal agents, and enhanced gaming. These features unlock a variety of edge computing applications that don't rely on cloud connectivity.
This technology has been validated across five generations of silicon. The EN100 represents the first commercially available product based on this technology, and the company has already secured multiple OEM partnerships. The underlying research came from Princeton University and has received over 144 million dollars in funding.
The EN100 chip consumes about as much power as a light bulb when running sophisticated AI models. It delivers 20 times better performance per watt compared to cloud and GPU alternatives, while also cutting CO2 emissions from cloud-based AI processing by a factor of 100. The chip achieves a computing density of 30 TOPS per square millimeter.
EnCharge AI has created breakthrough analog in-memory computing technology with its EN100 accelerator, delivering 20 times greater energy efficiency along with GPU-level performance for edge devices. It's well-suited for on-device AI inference where power consumption and physical size are major constraints that limit traditional solutions. As an early-stage hardware startup, the company has achieved significant technical validation but still needs OEM partnerships to reach the market.
The EnCharge EN100 is an efficient analog in-memory AI accelerator transforming the field of edge AI for OEMs creating the next generation of laptops, workstations, and AI PCs.
EnCharge AI has developed the analog in-memory AI accelerator EN100, which provides over 200 TOPS for laptops or 1 PetaOP/S for workstations and is 20 times more energy-efficient than competitors. The technology was proven through five silicon generations from a Princeton research spinoff company. Over $144 million has been invested into the company, with the goal of reaching the AI PC/Edge markets with an OEM early access program.
Good - comprehensive technical specs from multiple press releases and company technology page. No pricing, customer case studies, or API details publicly available (hardware startup focus).
Received $100M Oversubscribed Series B in February 2025, Bringing Total To $144M. Was spun out from Princeton University in 2022. 66 Employees Focused On AI PC And Edge Markets.
Has Developed Robust Analog Charge-Domain Computing Across Five Silicon Generations With A Variety Of Process Nodes. CEO Naveen Verma Stresses A Shift From Digital Limits To Scalable Inference Solutions.
100X Lower CO2 Emissions Than Alternatives That Are Cloud- and GPU-Based Through Processing On-Device. Supports ESG Objectives For Responsible AI Deployment By Reducing Power Consumption And Water Usage.
Round 1 Early Access Full; Round 2 Signup At Env1.EnchargeAI.com. Partners Developing Always-On Multimodal AI Agents And Real-Time Gaming Applications.
A Custom Built Suite With Optimization Tools, Compilation, And Support For PyTorch And TensorFlow. Can Handle Generative Models, Vision Transformers, CNNs Up To 128 GB LPDDR, 272 GB/s Bandwidth.
| Architecture Type | Primary Optimization | Framework Support | Scalability Model | Training vs Inference | Use Case Fit |
|---|---|---|---|---|---|
| GPU (NVIDIA/AMD) | General-purpose parallel compute | CUDA, PyTorch, TensorFlow, JAX | NVLink/InfiniBand clusters | Both (training dominant) | Data center mixed workloads |
| EnCharge EN100 (Analog IMC) | Analog in-memory matrix multiplication | Programmable AI frameworks | PCIe/M.2 multi-card | Inference-optimized | Edge devices, laptops, workstations |
| TPU (Google) | Systolic array matrix operations | TensorFlow, JAX (XLA) | Synchronized pods | Both (inference-optimized v7) | Cloud-scale training/inference |
| Accelerator | Primary Precisions | Quantization Support | Performance | Efficiency Gain | Target Workloads |
|---|---|---|---|---|---|
| EnCharge EN100 (Analog IMC) | INT8 (200+ TOPS) | Analog matrix multiplication | 200+ TOPS (M.2), 1 PetaOPS (PCIe) | 20x vs GPU competitors | On-device LLM inference, computer vision |
| NVIDIA H100 (Reference) | FP8, BF16, FP16 | FP8 quantization | ~4000 TFLOPS FP8 | Baseline | Data center training/inference |
| Workload Type | Optimal Hardware | Memory Requirement | Bandwidth Priority | Scaling Method | Key Metrics |
|---|---|---|---|---|---|
| Generative Language Models (On-Device) | EN100 M.2 / PCIe | Up to 128GB LPDDR | High (272 GB/s) | Multi-card PCIe | Tokens/sec, latency, energy efficiency |
| Real-Time Computer Vision | EN100 M.2 Laptop | Model size dependent | High (analog IMC) | Single card sufficient | FPS, power consumption |
| Professional AI Workstations | EN100 PCIe (4 NPUs) | 128GB+ datasets | Critical for large models | PCIe expansion | Throughput, cost efficiency |
