Insider Brief
- Sygaldry Technologies, a startup founded by quantum veterans Chad Rigetti and Idalia Friedson, is developing hybrid quantum-accelerated AI servers to address AI’s energy and performance bottlenecks.
- The company’s approach integrates multiple qubit modalities within a fault-tolerant architecture, aiming to enhance tasks like model training, inference and token generation in AI workloads.
- While still early-stage, Sygaldry’s focus on AI-specific quantum integration offers an approach that may help reduce data center energy demands.
A new startup from two veterans of the quantum computing industry is betting that quantum processors could ease AI’s growing energy and hardware burden by speeding up performance and reducing costs in ways GPUs can’t.
Sygaldry Technologies, launched by Rigetti Computing founder Chad Rigetti and former Strangeworks Chief Strategy Officer Idalia Friedson, is building what it calls quantum-accelerated AI servers, according to Bloomberg TV and social media posts. The goal is to combine classical infrastructure with quantum processors in a way that gives AI developers faster training, faster inference and better efficiency for models ranging from large language models to diffusion networks.
The startup is currently in Y Combinator’s Spring 2025 cohort and is now positioning itself as a hardware and systems company that is taking on two of the world’s biggest technical challenges: tackling the physical limitations of AI’s growth and finding ways to fully realize the potential of quantum computing.
“We’re addressing the problem of rising compute costs and energy bottlenecks in AI, which is on an unsustainable trajectory,” Friedson said. “The progress in quantum hardware since even just a few years ago is incredible, and this use case is crystal clear. We’re excited to work with our partners on this journey.”
AI Energy Demand
AI’s increasing power demands are pushing data centers and chipmakers to their limits, the company reports in its launch video. Running and training models, like GPT-4, requires hundreds of GPUs and massive energy consumption, a trend that analysts say is scaling faster than existing infrastructure can sustainably support. Sygaldry’s strategy is to tap quantum mechanics to reduce those hardware requirements — not by replacing classical infrastructure entirely, but by complementing it.
The company’s approach centers on a hybrid architecture that incorporates more than one type of qubit. Different qubit modalities — such as superconducting, trapped ion, or photonic — have strengths and weaknesses depending on the task. By combining these within a single system, the Sygaldry team hopes to create a fault-tolerant environment that improves reliability while exploiting the speedups quantum offers for certain types of linear algebra and optimization problems common in AI.
“Sygaldry quantum-accelerated AI servers combine multiple complementary qubit types within a fault-tolerant system architecture. By doing this, we are able to more easily meet the overall system requirements for utility-scale quantum computing,” said Rigetti. “We think this approach to quantum computer architecture will seem obvious in retrospect.”
Designed For High-Performance AI
According to the company, its quantum-accelerated servers aim to reduce model development cycles, enable quicker fine-tuning of existing models and increase the affordability of deploying high-performance AI. Sygaldry also reports its systems will allow for faster token generation, which is critical for deploying chatbots and virtual assistants, and faster inference in diffusion models, the core engine behind many AI-generated images.
Rigetti and Friedson bring deep expertise to the venture. Rigetti previously led Rigetti Computing from its inception in Y Combinator’s 2014 class through its IPO on the Nasdaq. Friedson served as Chief Strategy Officer at Strangeworks and earlier founded the first quantum policy center at the Hudson Institute. The pair also worked together to take Rigetti public in 2022.
Sygaldry’s focus on purpose-built integration with AI applications is unique, contrasting with a typical approach of building general-purpose quantum computers. While many quantum firms remain focused on solving problems, such as ones in chemistry or cryptography, Sygaldry is explicitly targeting the AI workflow. It plans to offer tools that allow AI researchers to incorporate quantum techniques into their existing pipelines, potentially making quantum acceleration as accessible as plugging in a new server.
The company is positioning its systems as a way to not only increase AI capabilities, but also to help offset AI’s environmental toll. Data centers already account for a growing amount of global electricity use, and projections suggest that number could rise steeply as AI adoption accelerates. By offering greater compute per watt, Sygaldry’s systems may act as a datacenter-level accelerator for the ongoing infrastructure expansion to support AI, according to the founders.
If successful, the company will be helping the quantum industry face down several challenges. Hybrid quantum-classical computing remains an experimental frontier, and many qubit modalities still struggle with error rates and stability.
While fault-tolerant quantum computing is a long-term goal across the field, some believe the technology is close to commercial maturity. Integrating multiple qubit types into one system may also add further complexity. Sygaldry’s servers will need to show measurable performance gains on real AI workloads — at a price that makes sense for customers.
The company has yet to disclose a product timeline or funding details.
0 Comments