Unconventional AI has reached an agreement to acquire Israeli artificial intelligence infrastructure company NeuReality, bringing its chip design expertise and AI inference technology into an ambitious effort to dramatically reduce the computing power and energy required to run AI systems.
The companies announced the definitive agreement Monday, saying the acquisition would accelerate Unconventional AI's development of a new computing architecture designed to deliver a 1,000-fold improvement in efficiency compared with conventional systems.
Financial terms of the transaction were not disclosed.
NeuReality, headquartered in Caesarea, Israel, specializes in AI inference infrastructure — the hardware and software systems used to run trained artificial intelligence models and generate responses to user requests.
Founded to address inefficiencies in conventional AI infrastructure, the company has spent seven years developing technology intended to reduce the computational overhead associated with coordinating AI workloads and moving data between components.
The acquisition will bring NeuReality's engineering team, including specialists in system architecture, semiconductor design and AI software, into Unconventional AI.
Targeting a 1,000-fold improvement in efficiency
Unconventional AI is developing what it describes as a four-dimensional computing substrate, an approach that seeks to improve AI processing efficiency by redesigning how computing hardware and AI models work together.
The company says achieving its target of a 1,000-fold efficiency improvement requires coordinated development across device physics, silicon architecture and AI models rather than incremental upgrades to existing chips.
The performance target remains a company objective, not a demonstrated result.
Unconventional AI said NeuReality's experience designing and deploying complete AI data center infrastructure would help shorten the development timeline for its next-generation computing systems.
"The hardest thing in this industry is assembling a team that can build real silicon and the systems around it, and Moshe has built one of the best," said Naveen Rao, co-founder and CEO of Unconventional AI, referring to NeuReality CEO Moshe Tanach.
"We are chasing a 1,000x improvement in AI compute efficiency, and that is a systems problem as much as a physics problem," Rao said.
"We think unconventionally about computing, and the accelerating pace of our research calls for an unconventional approach to growth. Bringing NeuReality's team into Unconventional accelerates our roadmap significantly."
The deal comes as AI companies face growing pressure to improve the economics of deploying increasingly sophisticated models.
While much of the industry's attention has focused on the computing resources required to train AI systems, inference — the process of running those models after training — also demands substantial infrastructure and energy.
NeuReality has focused on reducing the processing and data-movement overhead that contributes to those costs.
Israeli engineering team to join Unconventional AI
Tanach, NeuReality's co-founder and CEO, said the acquisition would allow the company to pursue its original mission on a larger scale.
"We built NeuReality because we believed the economics of AI inference were fundamentally broken, and that the answer had to come from rethinking the system rather than tuning it," Tanach said.
"Unconventional is taking that conviction further than anyone else in the market," he added. "With demand for AI computation growing against finite energy and semiconductor capacity, this mission has never been more essential."
Tanach described the acquisition as a continuation of NeuReality's efforts to improve the cost and efficiency of AI infrastructure.
"This is the next chapter of the mission we started with: fixing the economics of AI, now with the platform to do it at full scale," he said.
Under the agreement, Tanach and NeuReality's founding team will join Unconventional AI.
The acquiring company also intends to maintain NeuReality's engineering operations in Israel, preserving the company's local research and development presence.
The transaction has been approved by the boards of directors of both companies and is expected to close in the fourth quarter of 2026, subject to customary closing conditions.
The companies did not disclose the purchase price, the number of employees joining Unconventional AI or a specific timetable for bringing the proposed computing architecture into commercial production.


