Intactis Bio has launched Biostack , a head-to-head Tetris-like game that pits people against brain cells. The game is an interactive public demonstration of Intactis’s underlying tech, which Founder and CEO Daniel Rodriguez-Granrose, PhD, refers to as a biohybrid computer.

Intactis’s Biohybrid Processing Units (BPUs) are designed to fit into standard data center computing racks, but they contain more than just silicon. Inside each BPU are neurons derived from human stem cells, “wrapped in the cooling, life support, and signal hardware needed to keep it healthy and readable while it computes.”

The company claims that the living substrate in their BPUs is up to three million times more energy‑efficient per decision than silicon at scale, with projected total‑cost savings of 90 percent.

“This is about the massive energy savings of biohybrid systems over pure silicon systems,” Rodriguez-Granrose told me. “This game offers a tactile example of how this works, where people can really compete against these biohybrid machines and understand their value.”

Intactis has spent over two years designing the system, stimulating and optimizing neural responses from over 150 different interactions. This requires far greater complexity than other simple gaming demos with limited axes of control. The BPUs performance in the game represents inference, according to Rodriguez-Granrose, demonstrating how Intactis’ biotransformer model responds to a broad array of inputs, from characters like letters and numbers, to equations, or even multimedia.

Today, they are offering cloud-based access to their biocomputing capabilities, initially targeting customers that are already spending upwards of $20,000 a month on compute. The company plans to test their bold efficiency claims at scale by bringing their BPU to data center partners in the near future.

Rodriguez-Granrose is a former National Science Foundation Fellow who leads a team of 13 employees in Salt Lake City. This includes Tim Cloutier, PhD, who is Co-Founder and Chief Commercial Officer of Intactis Bio, leading commercial strategy and investor relations for the company. The company operates out of Altitude Labs, an accelerator run by Recursion Pharmaceuticals. Since their founding in 2024, the company has raised $1 million in capital from RPV, Convoi, and various grants, including Shift Grants and the state of Utah.

The field of biological computing was seeded twenty years ago by the 2006 discovery of induced pluripotent stem cells, which won a 2012 Nobel Prize on the way to kickstarting new industrialization efforts worldwide.

Today’s nascent industry is built on a rapidly expanding technical foundation of research and development across wetware methods, organoid intelligence and biological neural networks. Intactis joins a handful of startups building pick-and-shovel plays for different kinds of AI researchers around the world. A brief timeline of the global landscape:

  • 2014 - FinalSpark, founded by Fred Jordan and Marin Kutter in Switzerland, built the world’s first cloud-based biocomputing “Neuroplatform” for ten years before launching it publicly in 2024 , with nine pro-bono research partnerships. Their system uses 16 synthetically grown neural organoids consisting of about 10,000 neurons each. Today FinalSpark offers researchers free access to over 30 TB of recorded neuronal activity data, and a tier of paid subscriptions to run custom experiments fully remotely.
  • 2019 - Cortical Labs, founded by Hon Chong and Brett Kagan in Australia, has set out to deliver the world’s first commercial biocomputer. After demonstrating neurons playing Pong in 2022, the company launched their system, the CL1, in 2025. In 2026 they opened API-based access, shared a demo of their system playing Doom , and announced two new data centers including a multi-phase deployment in Singapore. Following a $10m seed in 2023, Cortical Labs has secured public grant support and raised additional capital to drive global growth.
  • 2022 - The Biological Computing Company was founded in Baltimore by Alex Ksendzovsky and Jonathan Pomeraniec, two neurosurgeons. The company emerged from stealth this February with $25m in funding and new San Francisco headquarters that houses a team of two dozen scientists and engineers hired from top tech firms and global research labs. They are building software adapters to improve performance of specific AI models, such as video generation , with commercial launch of an API-based marketplace planned later this year.

There are additional companies, including those operating in stealth today. More startups are likely to emerge in the years ahead thanks to a busy research ecosystem and a budding commercial supply chain. MaxWell Biosystems, a ten-year old German manufacturer of life sciences research systems, offers biocompute tools directly to hundreds of researchers and industry partners. They supplied the microelectrode arrays used in Cortical Labs’s past publications. Other vendors specialize in various biology, hardware, software layers.

From physical hardware to virtual clouds, with buy, build, or lease options available, today’s crop of startups represents a cottage industry, each advancing a different business model. Beyond distinct products and revenue strategies, each company uses unique terminology, different technology and science stacks, best practices, along with different approaches to governance and ethical oversight of their living biology programs.

Market Outlook: More Than Meets The Eye

There are a few ways to explore the opportunity presented by today’s biocompute startups.

The backdrop is today’s surging AI energy demands, which are on track to double or more by the end of the decade, with growing implications for the economy and the environment. Measured in terawatt hours (TWh), today’s annual global AI consumption of 400-500 TWh could cross 1,000 TWh by 2030. For context, 1 TWh is equivalent to 1,000 gigawatts or one million megawatts of energy.

In even the best case scenario, biocomputing efficiency will be one layer of the solution to the looming energy crisis, alongside new chip design, more efficient deployment , and other innovations. For context, Singapore’s data center project is a public effort to create 200MW of new capacity through more eco-friendly approaches; Cortical Labs’ deployment will start with a single rack of 20 units, which will be evaluated and expanded by up to 50x over the multi-year contract.

Experts argue that energy-efficiency claims from biocomputing are unfounded without a basic public audit of total energy required to build, maintain, and deliver a commercial-scale biocomputing operation. Other sought-after benefits of biological substrates over silicon, such as faster learning or better memory will also require more nuanced appraisals, in light of the limited lifecycles and replacement needs for biological computing units.

Ultimately, the early wave of customers interested in piloting biocompute will determine whether the value justifies the technology at this stage. These buyers will be swayed more by each company’s go to market tactics, business models, customer workflow, than their technical differentiators or stated claims alone.

The energy demands of today’s industry are surging forward but shifting as AI evolves: The boom of building and training new models has expanded into live deployments running real-time applications running inference-based computing to operate in the real world. Moreover, general-purpose large language models (LLMs) built on graphical processing units (GPU) have proliferated into deeper, specialized use cases.

Investors at Fusion Fund estimate that today’s compute demand is 80% for inference and 20% for training, a full inversion from just two years ago. This is creating a post-GPU shift towards specialized chips and custom deployments. Spending patterns are changing too, from larger, timebound contracts to an ongoing cost structure.

Today’s early adopter market for biocompute-as-a-service looks to be AI-forward companies burning cash on compute while to develop specialized applications across gaming, video, robotics, drones, sensors, vision, and other segments across deep tech, biotech, and beyond.

Thesis-Driven Frontier Bets

Beyond the energy-savings play, biocomputing startups represent a promising, investor-backable evolution of next-gen AI R&D.

The opportunity transcends building software. For example, a recent biocomputing project funded by DARPA seeks to advance computing applications of interest to the US military. Some investors are bullish on backing the companies defining what they see is an inevitable shift in how intelligence is understood and constructed.

As per Fusion Fund’s memo: “This is the defining tension of the decade. The widening gap between what AI asks of silicon and what silicon can return is the single most important unlock in the technology landscape.”

Flourish Labs just emerged with $500m in funding led by Jeff Bezos to develop better AI using biology as ground truth to build better software and/or chips in the next few years. Lux Capital noted that they backed “the contrarian thesis that begins where the AI consensus ends: the data center capex curve is unsustainable, and thus the future of intelligence is not centralized but distributed, not detached but embodied, not trained once and frozen but continuously learning at the edge.”

As Intactis angles for position in the early biocompute market, a growing global audience will be keeping score.

Biostack is now available at play.intactis.bio. All details on session release timing, playing auction bids, and beta testing signup are available online.