Intactis

Intactis Bio Corp. is a Salt Lake City, Utah biocomputation company developing the Biohybrid Processing Unit (BPU), a rack-mountable AI accelerator that couples silicon processors with living, lab-grown neural tissue via a brain-machine interface. The platform encodes datasets as patterns of electrical pulses, delivers them to neural tissue through multi-electrode arrays, and uses reward-based and biochemical feedback to train the tissue to solve problems; outputs are recorded and decoded so the system can integrate with conventional AI/ML models while also exploring solutions through biological dynamics. The company frames the opportunity around the energy ceiling of silicon-based AI, stating that on current trends silicon AI power demand would exceed global electricity output by 2040, and positions its biocomputer for co-location with energy-constrained data centers running AI/ML and high-performance workloads.

The technical approach starts with induced pluripotent stem cells (in one account derived from human blood samples) differentiated into neurons, which are 3D-bioprinted and vascularized into three-dimensional substrates sustained in custom brain-machine bioreactors. Each substrate is described as containing tens of millions of neurons, approximately the size of a grade-school eraser and reported as 600,000 times larger than previously demonstrated comparable engineered neural tissue, with a scalable bioprocess intended to keep centimeter-scale tissue viable long term. A proof-of-concept experiment reported neurons controlling their own environment by determining when to pulse their bioreactor with nutrients over 21 days, acting as sensor, controller and computational engine; the work has been patented in multiple countries (reported as more than seven) but was not peer-reviewed at the time of reporting. In March 2026 the company announced a demonstration in which iPSC-derived neurons coupled to silicon performed arithmetic operations and transmitted the phrase "Hello, World!".

A software layer called the Biotransformer serves as a plug-and-play interface between living neurons and silicon systems, integrating with PyTorch so developers can use biological computation in standard AI workflows without neuroscience expertise. Named target verticals include gaming (adaptive AI opponents), large language models (adaptive reasoning layers) and TechBio/life sciences discovery pipelines. Intactis projects a 95% reduction in energy costs, a 90% reduction in total costs and an 88% reduction in data center footprint relative to exaflop-scale silicon systems, with a long-term goal of exaflop-scale biohybrid machines.

Founding story

The company was founded in 2024 by Daniel Rodriguez-Granrose, who holds a PhD and PSM and has roughly 14-15 years of experience in cell and tissue engineering and AI/ML, was an NSF Fellow, and led the first team to scale neural organoids into 100L+ bioreactors. He has said the combination of having scaled neural cultures in large bioreactors and observing the AI energy crisis made it clear that building AI on a neural tissue substrate rather than silicon chips could deliver both computational efficiency and reduced environmental impact. He self-funded the company with $150,000.

Business model

B2B. The company offers subsidized AI/ML data-analysis services as part of its biocomputation alpha, pairing paid analysis using traditional methods with free analysis on the biocomputation platform, and invites customers to build AI/ML workloads on its biological platform. It has launched a Phase 1 Biocomputation Alpha with free and paid tiers, where paid users receive priority capacity and co-development access; Phase 2 is planned for 2027. Its model emphasizes partnerships and collaborations with research institutions, biotechnology firms and data-centric companies, and co-location of its biocomputer with energy-constrained data centers.

Revenue is described as coming from subsidized AI/ML analysis services tied to the biocomputation platform, including paid tiers of the Phase 1 alpha with prioritized capacity and co-development access, and from partnership-based, customized engagements within a B2B framework.

Traction

In March 2026 the company reported that lab-grown iPSC-derived neurons coupled to silicon performed arithmetic operations and transmitted "Hello, World!". An earlier proof-of-concept had neurons regulating nutrient pulsing of their own bioreactor over 21 days. The Phase 1 Biocomputation Alpha has launched with free and paid tiers, with Phase 2 planned for 2027. The company reports about 6 employees, a team spanning neuroscience, hardware engineering and commercialization, and patents granted in multiple countries. It won its group competition at the BioUtah Entrepreneur & Investor Summit and is a BioUtah Standard Community member.

Latest developments

March 2026: announcement of arithmetic operations and a "Hello, World!" transmission by lab-grown neurons interfaced with silicon, together with a $250,000 investment from Nucleus Fund intended to expand the platform's computational complexity, plus a SHIFT grant and an Altitude Lab Trail Angel Grant. The Phase 1 Biocomputation Alpha is live with free and paid tiers, Phase 2 is slated for 2027, and the $2.5 million seed round remains open. The company is recruiting computational neuroscientists, iPSC cell culture staff and AI/ML engineers.