Wetware as a Service: The Strategic Case for Renting Living Brains

Wetware as a Service: The Strategic Case for Renting Living Brains

By M. Mahmood, Strategist and Consultant

TL;DR / Summary

Every generation of enterprise technology eventually confronts the same uncomfortable question, which is whether the substrate powering the business can keep pace with the ambition of the business itself. Artificial intelligence (AI) has now reached that inflection point. Silicon compute has become so expensive and so energy-hungry that Anthropic alone may spend $750 billion on computing capacity while reporting a net loss exceeding $80 billion for the first half of 2026, and the IMF has openly warned that the AI investment bubble could destabilize the global economy if returns fail to materialize.

Against that backdrop, a radical alternative is quietly moving from laboratory curiosity to commercial service. Wetware as a service, often shortened to WaaS, allows organizations to rent remote access to living neural tissue the way they once rented GPU time, and the implications for cost structures, capability development, and competitive positioning deserve far more executive attention than they currently receive.

Key Takeaways

• The silicon strain is real: Anthropic's projected $750 billion compute spend and the IMF's AI bubble warning underscore why energy-efficient alternatives now matter to boardrooms.
• The service model is live: Cortical Labs rents access to its CL1 living-neuron systems (roughly $35,000 hardware, or cloud API access), and FinalSpark offers remote access to brain organoids for research, making experimentation available at ordinary research budgets.
• Biology's edge is energy and learning efficiency: Neurons operate on watts rather than megawatts and generalize from few examples, proven by demos like Braille recognition and a Doom-playing neural colony.
• Value pools are specific, not general: Drug discovery, adaptive robotics, optimization under uncertainty, and biosensing are where WaaS will land first, not general-purpose computing.
• Governance can't wait: Donor consent, organoid sentience debates, and living-tissue continuity risks demand frontier-safety-style frameworks with named owners before scaling.
• The recommended posture: Run a low-cost pilot now, build defensibility through data and workflow ownership, and preserve option value against an energy-constrained future.

What Wetware as a Service Actually Delivers

A wetware computer replaces silicon transistors with living neurons, typically grown from human induced pluripotent stem cells and cultured on electrode arrays that translate electrical signals into machine-readable data. Because neurons learn, adapt, and self-organize rather than execute fixed instructions, these systems occupy a fundamentally different niche than conventional processors, offering the possibility of dramatically more energy-efficient computing for specific classes of problems.

The service model matters as much as the science here. As early analysis of the Wetware-as-a-Service model describes, WaaS abstracts away the immense complexity of cultivating and maintaining biological tissue, letting researchers and developers experiment with biological computation through familiar cloud interfaces such as APIs and web consoles, with pay-as-you-go billing that mirrors how enterprises already rent accelerator time.

Three organizations define the current market: 

1. Cortical Labs in Australia commercializes the CL1, a shoebox-sized unit that keeps roughly 800,000 lab-grown human neurons alive on a silicon chip and sells both the hardware and remote access to it. 
2. FinalSpark, a Swiss firm founded in 2014, operates an open and remotely accessible Neuroplatform for research that gives vetted institutions round-the-clock programmatic access to living brain organoids over a standard internet connection. 
3. Koniku, based in the United States, pursues a complementary path by fusing neurons with silicon to build biological sensing systems for security and diagnostics applications.

Why Biology Is Entering the Compute Conversation Now

Three forces are converging to make living compute commercially relevant for the first time.

The first is energy economics. The human brain performs staggering feats of learning and inference on roughly 20 watts, which is less power than a desk lamp, while modern AI data centers measure consumption in hundreds of megawatts, and researchers at FinalSpark estimate that biological computing could eventually operate up to a million times more efficiently than digital hardware on suitable workloads. When hyperscalers are projecting capital commitments in the hundreds of billions, even a niche technology with plausible efficiency advantages of several orders of magnitude deserves boardroom attention rather than dismissal.

The second force is learning efficiency. Biological neural networks generalize from remarkably few examples, which contrasts sharply with the data-hungry training regimes that dominate contemporary machine learning. In an especially striking demonstration from October 2026, computers made of human brain tissue learned to identify Braille letters, a result that researchers positioned within the emerging discipline of organoid intelligence, which treats lab-grown neural structures as genuine computational substrates rather than scientific curiosities. Earlier that same year, an independent developer with nothing more than Python access to rented neurons taught a colony of living brain cells to play the video game Doom in approximately one week.

The third force is market maturation. Cortical Labs now operates what journalists have described as a biological data center containing millions of living human brain cells that process information for paying customers, and the company offers its CL1 wetware systems either as a purchasable unit priced near $35,000 or through its rental model for organizations that want exposure without owning wet lab infrastructure. When a technology acquires a price list, a service-level conversation, and a developer community, it has crossed the threshold from science project to strategic option.

Where the Early Value Pools Are Forming

Executives evaluating WaaS should resist the temptation to view it as a general replacement for silicon, because the near-term value concentrates in workloads where biology holds a native advantage.

Drug discovery and neuropharmacology represent the most immediate commercial beachhead, since neurons derived from patients with specific conditions allow pharmaceutical teams to test compounds against living disease models and observe responses that neither animal testing nor digital simulation can reproduce faithfully.

Robotics and embodied control form a second attractive pool, because biological controllers handle noisy, unpredictable real-world feedback with a fluidity that rule-based and model-based digital systems struggle to match, and hybrid architectures that pair living adaptability with silicon precision may unlock new categories of autonomous machines.

Optimization under uncertainty offers a third opportunity, particularly for problems involving incomplete information, shifting constraints, and continuous adaptation, where the innate plasticity of neural tissue outperforms rigid algorithmic approaches that require constant retraining.

Biosensing and security screening round out the landscape, with Koniku pursuing what it describes as the full-stack merger of biological technology with applications that range from airport screening to agricultural monitoring and health diagnostics.

The Risk Agenda Leaders Cannot Delegate

Every transformative technology arrives carrying governance obligations, and wetware compounds the usual technology risks with entirely novel ethical dimensions that boards should confront directly rather than delegate to technical teams.

Consent and provenance sit at the top of the agenda, since these systems are built from cells that originated in human donors, and organizations experimenting with WaaS face legitimate questions about whether donors understood that their cellular material might eventually power commercial computation, as well as questions about privacy and identity that legal frameworks have barely begun to address.

The possibility of rudimentary sentience deserves equal seriousness, because scientists and ethicists have begun debating whether sufficiently complex organoids might develop primitive forms of experience, and while today's systems remain far from any plausible threshold of consciousness, prudent leaders will establish monitoring protocols and ethical guardrails now rather than scramble reactively if research advances faster than policy.

Operational resilience introduces a different category of exposure, given that living tissue dies, drifts genetically, and depends on life-support infrastructure whose failure modes look nothing like a server outage, which means continuity planning for biological compute demands fresh thinking about redundancy, vendor dependence, and graceful degradation.

Experience from the broader AI industry offers useful scaffolding here, since the discipline embodied in Google DeepMind's Frontier Safety Framework and OpenAI's work on safety cases for frontier model training demonstrates how pre-committed thresholds, named owners, and staged access tiers can govern emerging capabilities before incidents force the issue. Organizations entering wetware would benefit from applying similar rigor, and the structure of an effective frontier AI safety framework, which separates capability evidence, containment evidence, and commercial access into independent reviews with distinct accountable owners, translates remarkably well to biological computing pilots.

Five Moves for Forward-Looking Executives

Leaders who want optionality on this technology without premature commitment can act deliberately through five practical steps.

1. Map candidate workloads now, by identifying corners of the business where learning efficiency, energy constraints, or biological realism create pain that digital systems handle poorly, with neuropharma research, sensory analytics, and adaptive control problems serving as natural starting points.
2. Run a bounded pilot, using the rental access already available through FinalSpark's Neuroplatform or Cortical Labs' cloud offering, since the cost of a structured experiment with living neurons now sits within reach of an ordinary research budget rather than requiring dedicated wet lab investment.
3. Establish governance before scale, assigning named owners for ethical review, biosafety oversight, and vendor diligence, and treating any evidence of unexpected emergent behavior in neural substrates with the same seriousness that frontier AI labs now apply to model evaluations.
4. Build defensibility intentionally, because history suggests that winners in new compute paradigms own workflow position, proprietary outcome data, and customer trust rather than the underlying substrate itself, a logic explored in depth in the case for why AI startup defensibility matters more than speed, and the lessons translate directly to whatever application layer emerges atop wetware infrastructure.
5. Scenario-plan the energy-constrained future, since organizations that assume silicon costs and power availability will follow historical curves may find themselves strategically exposed if regulatory, environmental, or economic shocks accelerate the search for biological alternatives, and a small working group tracking WaaS milestones costs almost nothing relative to the option value it preserves.

The Bottom Line

Wetware as a service will not displace GPUs this decade, and executives should discount any vendor promising otherwise, but dismissing the field would repeat the error that incumbents made with cloud computing and then with generative AI. The strategic question facing leadership teams is not whether living computation will eventually occupy a meaningful niche in the enterprise stack, because the commercial evidence from Melbourne to Switzerland increasingly confirms that it will, but whether your organization will develop the judgment, relationships, and governance muscles to exploit that niche before competitors do. In a period when AI compute economics strain credibility and boardrooms worldwide are hunting for the next durable efficiency frontier, the companies that treat biology as a strategic resource rather than a novelty will shape the terms of whatever comes next, and the window for building that position has already opened.