EvoChip targets lower-cost, GPU-free AI inference via AltiCore software-defined platform.

EvoChip, a Dana Point, California-based startup, is launching AltiCore, a software ecosystem designed to dramatically reduce the power requirements and costs of AI inference. The technology represents a fundamental departure from conventional neural network approaches, instead using logic gates on standard processors to handle AI computations. According to Patrick O'Neill, the company's chief technology officer and co-founder, AltiCore is purpose-built for AI rather than adapting general-purpose computing infrastructure.

The core innovation replaces heavy arithmetic operations with logic-dominant operator chains. "We wanted to reinvent that stack from the ground up," O'Neill said. "We wanted something purpose-built for AI, not applying general-purpose compute towards AI, and in the process come up with an efficient way of reducing requirements for power, compute, memory, processing time, all that stuff."

O'Neill explains that the technology scales from microcontroller units (MCUs) to large server installations. Unlike traditional neural networks, which rely on complex matrices of weights, coefficients, and interconnections poorly suited to digital computing, AltiCore employs the fundamental building blocks of digital logic. "We went back to the fundamentals," he said. "We used basic building blocks of digital logic—logic gates, ANDs and Ors, and NANDs and NORs, and bit shifts and primitives that are very fast and lightweight in the digital silicon transistor substrate."

The efficiency gains are substantial. Conventional mathematical approaches require tens of thousands of logic gates; AltiCore uses only six. An AI model that normally requires a high-end server can now run on smaller chips like STMicroelectronics MPUs without modification.

EvoChip and Sidepath, a Laguna Hills, California-based Dell Technologies Titanium partner, tested the technology on Dell server and laptop hardware using a University of California, Irvine dataset for credit default prediction. The neural network comparison required 8,323 arithmetic operations per inference versus 116 for AltiCore. In a second test, the differential was 21,000 operations versus 110. When tested on a Dell notebook, AltiCore achieved higher inferences per second than a neural network achieved on a $60,000 server rack. While the results have not undergone third-party audit, O'Neill noted they are reproducible, with detailed test reports available for independent verification.

Patrick Mulvee, a Sidepath partner, called the technology a game-changer. Testing on both new Dell Pro Max laptops and older models, as well as the newest Dell R670 servers with and without virtual machines, demonstrated that "when we ran them using the EvoChip technology, they perform exponentially better, faster, quicker, and pull less power. It makes a server perform like it's on steroids." Critically, no hardware changes were required—only the addition of EvoChip software.

The technology has broad commercial applications. Matt Musial, Sidepath's solutions architect, emphasized that AltiCore significantly lowers entry barriers to AI adoption. "We could apply this technology to all sorts of things now because it doesn't have a $100,000 or a million-dollar starting point," Musial said. He noted that conventional approaches typically result in 95 percent project failure rates, whereas this technology enables rapid AI evaluation on standard hardware, allowing customers to "start on your laptop. You can start on a much smaller initial investment and then figure out what works, what you need in a model."

O'Neill acknowledged that the technology has not yet been applied to large-scale models. The immediate market focus is microcontrollers, where neural networks cannot fit at all. "Ultimately, we would like to replace Nvidia GPUs," he said. "If we can make this thing outperform a $3,000 GPU on a desktop, that's a real value in the marketplace." The potential benefit is substantial: if a single server running AltiCore can match the throughput of 27 servers using legacy algorithms, the savings compound across power, cooling, and infrastructure costs.

EvoChip is moving AltiCore toward general availability while building documentation, refining the user experience, and expanding platform support. Channel partners working with system vendors will find multiple value-add opportunities, particularly where reducing server count or power consumption directly impacts customer budgets and operational constraints.