Brain Cells Playing DOOM Signals a Biocomputing Era

What’s Happening

A video circulating widely this week shows something that sounds ripped from science fiction: living human brain cells — actual neurons grown in a lab — playing the classic 1993 first-person shooter DOOM on a CL1 bioprocessor. The demonstration, which has blown up on tech forums and social media, represents a dramatic leap from earlier proof-of-concept experiments where biological neurons were taught to interact with Pong. Now they’re navigating corridors, responding to enemies, and processing visual input in real time on a vastly more complex game. The implications go far beyond a party trick.

This builds on work pioneered by Cortical Labs, the Australian biotech startup that made global headlines in 2022 when it published research in Nature Electronics showing that a dish of 800,000 human brain cells — dubbed “DishBrain” — could learn to play Pong in five minutes. The CL1 is the company’s commercial-grade successor: a hardware-wetware hybrid platform that connects living neuronal cultures to silicon interfaces. The DOOM demo isn’t just showmanship. It’s a benchmark — proof that biological neural networks can handle inputs of a complexity that, frankly, most observers in the field didn’t expect to see this soon.

The timing matters. We’re watching the birth of organoid intelligence (OI), a term coined by researchers at Johns Hopkins University in early 2023 to describe computing architectures that use three-dimensional cultures of human brain cells as processing units. While traditional AI runs on GPUs burning megawatts of power, a biological neural network the size of a pinhead can perform certain learning tasks with almost negligible energy consumption. In a world where data center power demands are becoming a genuine infrastructure crisis, that fact alone makes this space strategically important.

Why Founders Should Care

Let’s get the obvious out of the way: nobody is going to replace your cloud servers with brain cells next quarter. But the underlying technology trend here — biological computing — is entering the window where enormous venture capital opportunity lives. That sweet spot between “this is just a research curiosity” and “this is an established market dominated by incumbents” is where fortunes get built. And multiple signals suggest we’re right in it.

First, the energy argument is staggering. Training a large language model like GPT-4 is estimated to have consumed roughly 50 gigawatt-hours of electricity. The human brain, which outperforms these models on most general intelligence tasks, runs on about 20 watts — the equivalent of a dim light bulb. If biocomputing can capture even a fraction of that efficiency advantage in narrow commercial applications, you’re looking at a market disruption comparable to the shift from mainframes to microprocessors. Research firm Allied Market Research projected the global neuromorphic computing market (the broader category that includes bio-inspired and biological architectures) could reach $21.3 billion by 2032.

Second, the regulatory and ethical landscape — while complex — is actively being shaped right now. The EU’s AI Act doesn’t directly address biological computing. The FDA doesn’t have a clear framework for non-therapeutic biological processors. This is the kind of gray zone where startups that engage early with regulators get to help write the rules, rather than being crushed by them later.

“We’re at the very beginning of a new kind of computing. The question isn’t whether biological processors will be useful — it’s which applications will they dominate first.” — Dr. Brett Kagan, Chief Scientific Officer, Cortical Labs

Third — and this is the practical signal for founders scanning for timing — DOOM has always been the tech community’s unofficial benchmark for “can this platform actually compute?” People have run DOOM on pregnancy tests, on tractors, on bacteria. But those were gimmicks, running pre-rendered outputs. The CL1 demo is different. The biological neurons are processing visual input and generating motor responses in a closed feedback loop. That’s real computation, and the fact that it works on a task this complex means the platform is approaching the threshold where commercial applications become feasible.

Who’s Already Moving

Cortical Labs (Melbourne, Australia) is the company behind the CL1 and the DishBrain research. They’ve raised over $10 million and are positioning themselves as the NVIDIA of biocomputing — building the foundational hardware-wetware platform that other companies will build on top of. Their business model centers on leasing access to biological processing units for drug discovery, AI research, and defense applications.

FinalSpark (Vevey, Switzerland) launched what it calls the world’s first “Neuroplatform” — a remote-access biocomputing platform where researchers can run experiments on living brain organoids over the internet. Think of it as AWS, but the processors are alive. They went live in 2024 and have been onboarding academic and commercial users since. Their claim: their biological processors use a million times less power than traditional silicon for comparable learning tasks.

Koniku (San Francisco, USA) has been working on biological-silicon hybrid chips since 2014, initially targeting scent detection and air quality monitoring for airports and defense installations. They’ve taken a more applied, vertical approach — using engineered neurons for specific sensory tasks rather than general computation. They reportedly secured over $10 million in contracts and funding.

Johns Hopkins Organoid Intelligence initiative isn’t a startup, but the academic consortium led by Dr. Thomas Hartung is setting the research agenda for the entire field. Any founder entering this space should be reading their papers and tracking their collaborators, because that’s where the next wave of spinouts will originate.

The Startup Opportunity

You don’t need to be a neuroscientist to build a company in this space. Here’s where the real whitespace is:

• Tooling and infrastructure. As biocomputing platforms like the CL1 and Neuroplatform mature, they’ll need developer tools, APIs, monitoring systems, and data pipelines. Think of the companies that built the picks-and-shovels around cloud computing — the equivalent layer for biocomputing barely exists yet. Somebody needs to build the “Datadog for organoid computing.”
• Drug discovery acceleration. Living neural networks offer an extraordinary opportunity for pharmaceutical companies that want to test compounds on functional human brain tissue without clinical trials. A startup that packages biocomputing as a drug screening service — particularly for neurological conditions like Alzheimer’s or Parkinson’s — could access a multi-billion-dollar pharma services market.
• Hybrid AI architectures. The future probably isn’t purely biological or purely silicon. There’s an opening for startups that develop frameworks for combining traditional machine learning with biological co-processors — letting the wet neurons handle tasks they’re naturally good at (pattern recognition, adaptation, low-energy learning) while silicon handles raw throughput.
• Ethical governance and certification. This one is underrated. As biocomputing scales, companies, governments, and the public will demand ethical standards. Who certifies that a biocomputer doesn’t experience pain? Who audits the sourcing of human cell lines? A startup that establishes itself as the trusted authority in biocomputing ethics could become essential infrastructure — think UL certification but for living processors.
• Education and simulation. Biocomputing is going to need thousands of trained operators and researchers. Platforms that simulate biological computing environments for training purposes, or that provide educational content, will find eager customers among universities and corporate R&D labs.

Watch Out For

Let’s be honest about the real obstacles, because this space has landmines.

Ethical backlash is the most obvious risk. The phrase “human brain cells” triggers deep discomfort for many people. It doesn’t matter that these organoids are nowhere near conscious — they’re tiny clusters of neurons with no sensory experience or awareness. The perception that companies are “growing brains in jars” will be weaponized by critics. Any founder in this space needs a proactive communications strategy and a genuine commitment to ethical guardrails, not just a page on their website.

The science is still early. DOOM is impressive, but biological neural networks remain unpredictable, fragile, and difficult to scale. Organoids degrade over time. Reproducibility is a challenge. If you’re building on this technology, you need long time horizons and patient capital — this isn’t a space for founders who need revenue in 12 months.

Regulatory uncertainty cuts both ways. Yes, early entrants can help shape regulations. But they can also get crushed by sudden, poorly designed rules. The moment a tabloid runs a “Scientists grow thinking brains in lab” headline and politicians react, you could see heavy-handed restrictions that freeze the entire sector. Founders should be actively engaging with policymakers, not hoping they’ll be ignored.

IP and cell line sourcing. Where do the human cells come from? iPSC (induced pluripotent stem cell) lines are the standard, but sourcing, consent, and IP rights around specific cell lines are legally complex. Getting this wrong could mean existential lawsuits.

The hype cycle is real. Every time a biocomputing demo goes viral, expectations inflate. If commercial applications take longer than investors expect — and they usually do in deep tech — the funding environment could cool dramatically. Build with the assumption that your next round might be harder to close than your last.

The Bottom Line

Living brain cells playing DOOM isn’t just a wild internet video — it’s a timestamp. It marks the moment biological computing crossed from pure research into demonstrable, functional computation on a recognizable benchmark. For founders with the right technical chops or domain expertise, this is a once-in-a-generation platform shift in its earliest innings. The infrastructure layer is wide open, the application verticals (pharma, defense, AI) are massive, and the incumbents haven’t woken up yet.

But this isn’t a space for the faint of heart or the short of runway. The science is young, the ethics are complicated, and the public narrative could turn hostile overnight. If you’re going to build here, come prepared: deep partnerships with research labs, a credible ethical framework from day one, and investors who understand that deep tech timelines aren’t SaaS timelines. The founders who get this right won’t just build companies — they’ll define an entirely new category of computing. And that’s a bet worth studying seriously over your next cup of coffee.