The Download: mice with part-human brains and climate tech innovators

When I first launched WhatIsFuture.com, my mission was to cut through the noise and evaluate whether emerging technologies were truly revolutionary or merely expensive marketing campaigns. Over the past few months, I have been watching the artificial intelligence land-grab with a mix of fascination and profound alarm. If you listen to the echo chamber of venture capitalists and hardware vendors, you would think the path to Artificial General Intelligence (AGI) is simple: construct larger data centers, purchase hundreds of thousands of high-end graphics processors, and hook them up directly to electrical sub-stations. But as I analyze the trajectory of this industry, I have come to a stark realization that I cannot ignore.

Silicon Valley is currently trapped in a massive collective delusion. We are spending hundreds of billions of dollars trying to brute-force machine intelligence by dumping megawatts of power into towering stacks of hot GPUs. We are treating intelligence as a pure engineering problem of compute power and data size. But every now and then, a breakthrough occurs that shatters our tidy assumptions about hardware, biology, and energy. Two recent, seemingly disparate headlines captured my attention: neuroscientists successfully integrating human brain organoids into living rodent brains to process information, and a new generation of climate tech innovators racing to keep our melting power grids alive under the weight of AI workloads. When you step back and view these developments together, the signal in the noise becomes crystal clear.

The Silicon Trap: The Thermodynamic Crisis of Brute-Force AI

In my view, the current architecture of artificial intelligence is fundamentally unsustainable. Today's state-of-the-art Large Language Models (LLMs) require massive cluster deployments running tens of thousands of specialized chips for months on end just to complete a single training run. The electrical consumption of these facilities is staggering. I have talked with grid operators who are deeply concerned about how to supply gigawatts of power to planned data center campuses without destabilizing entire regional energy grids.

Consider the stark physical realities that we face today:

• Astronomical Energy Draw: A top-tier AI training run can consume tens of gigawatt-hours of electricity, equivalent to the power consumption of thousands of residential homes over an entire year.
• Thermal Dissipation Limits: As chip designers push transistor densities to physical limits, cooling these systems requires hundreds of millions of gallons of water and complex cooling infrastructure.
• Diminishing Returns: Scaling laws suggest that to achieve marginal improvements in model capability, we require exponential increases in data, compute, and electricity.

We are building digital furnaces to emulate a fraction of what nature achieves on a fraction of the energy budget. This is why major tech enterprises are now buying up nuclear energy rights and investing directly in experimental power generation. In my opinion, attempting to power a digital brain by burning gigawatts of carbon-heavy or grid-straining energy is an inefficient paradigm. We are hitting a physical wall, and the hardware industry knows it.

Mice with Part-Human Brains: The Dawn of Biological Computing

While the tech sector is busy pouring concrete for data centers, bio-engineers are quietly approaching the problem from the opposite direction. Scientists have successfully grown three-dimensional clusters of human brain cells—known as human cerebral organoids—and implanted them into the brains of newborn mice. These are not merely passive grafts. The human neurons integrate directly into the host rodent's vascular system, receive blood flow, form functional synaptic connections with mouse brain tissue, and respond to sensory stimuli.

The human brain operates on roughly 20 watts of power—barely enough to light a dim dimmable bulb—yet it performs complex reasoning, spatial navigation, emotional processing, and real-time learning that our largest server farms cannot replicate.

When I look at experiments involving organoid intelligence and hybrid bio-silicon systems, I see the early blueprint for a radical hardware revolution: wetware computing. Research groups and startups are already connecting these living neural cultures to microelectrode arrays, teaching biological neural networks to play video games, process spatial data, and perform pattern recognition tasks at a fraction of a fraction of a single watt.

Why is this so disruptive? Because biological systems evolved to solve the exact bottlenecks that currently plague our silicon systems:

• Extreme Energy Efficiency: Living neurons compute using chemical gradients and ion channels, operating near physical limits of energy efficiency.
• Parallel Plasticity: Human neural networks continuously rewire their physical structure based on experience, enabling true continuous learning without expensive backpropagation passes.
• Fault Tolerance: Biological brains gracefully handle cell death and signal noise without catastrophic system failure.

I am convinced that the future of computing will not belong solely to silicon or biology, but to hybrid platforms that combine the precise speed of digital hardware with the extreme efficiency of biological networks.

The Climate Tech Rescueline: Bridging the Gap

Until bio-computing matures beyond laboratory benchtop experiments, the tech sector must deal with the immediate reality of an unprecedented energy crunch. This is where climate tech innovators enter the picture. Over the past two years, I have seen a remarkable shift in how clean-energy founders frame their mission. Climate tech is no longer just about meeting compliance standards or hitting zero-emission goals; it has become the essential engine keeping advanced technology alive.

I have been closely monitoring three key climate tech sectors that are stepping up to power the future of computing:

1. Next-Generation Geothermal Energy

Traditional geothermal energy was restricted to geologically active regions like Iceland or California. However, advanced geothermal startups are using directional drilling techniques borrowed from the oil and gas industry to harvest deep earth heat anywhere on the planet. This provides clean, continuous baseload power that does not fluctuate with weather conditions—making it ideal for power-hungry compute facilities.

2. Advanced Nuclear and Micro-Reactors

Small Modular Reactors (SMRs) and micro-reactors are transitioning from concept to reality. Tech leaders are forming direct partnerships with nuclear energy providers to build dedicated clean-power sources next to modern data centers, bypassing traditional utility transmission bottlenecks entirely.

3. AI-Driven Smart Grid Management

Ironically, software models are being deployed to optimize energy delivery across aging electrical grids. By using predictive analytics, these systems can balance intermittent renewable inputs like solar and wind, dynamic data center workloads, and municipal energy demand in real time to prevent blackouts.

In my view, this interaction between energy supply and compute demand is one of the most compelling trends of our generation. The high power demands of AI are forcing rapid capital deployment into clean energy infrastructure that might otherwise have taken decades to secure funding.

The Ethical and Technological Horizon

As the line between living biological tissue and mechanical computing continues to blur, we must confront deep ethical questions. Integrating human brain tissue into animal hosts or wiring living neural organoids into synthetic feedback loops raises serious moral challenges. Where do we draw the line between a biological array and a sentient entity? What protections apply to hybrid biological systems once they demonstrate self-organizing capabilities?

At WhatIsFuture.com, my goal is to track these questions before they become crises. I do not believe we should halt biological computing research out of fear, but we must establish clear ethical boundaries before these systems scale commercially. At the same time, we must hold the hardware industry accountable for its energy consumption. We cannot allow the pursuit of digital intelligence to undermine environmental sustainability.

We stand at a unique crossroad. On one path lies the brute-force route: burning gigawatts of electricity on silicon architectures that are hitting physical limits. On the other lies a multi-disciplinary frontier that combines synthetic biology, advanced climate technologies, and bio-hybrid architecture. I know which vision of the future I am betting on.

Frequently Asked Questions

What are "part-human brain mice" and why are scientists creating them?

These are lab mice that have had human brain organoids—small, lab-grown 3D clusters of human brain cells—transplanted into their cerebral cortexes as infants. Scientists build these hybrid models to study human brain development, neurological diseases, and functional neural integration under natural living conditions. These models also help researchers understand how biological neural networks efficiently form connections and process information.

How does biological computing compare to traditional GPU clusters in energy efficiency?

Biological computing, or wetware, is vastly more energy efficient than silicon chips. The human brain consumes roughly 20 watts of power while performing complex real-time cognitive tasks. In contrast, modern AI data centers require megawatts to gigawatts of electrical power to process equivalent complex tasks using brute-force matrix operations. Biological systems compute using chemical energy and adaptive physical rewiring, which uses a tiny fraction of the power required by digital transistors.

Can climate tech solve the energy crisis caused by modern AI models?

Climate tech innovations—such as advanced geothermal energy, Small Modular Reactors (SMRs), and smart grid software—are essential for meeting short-term energy demands cleanly. However, generation tech alone cannot keep up with exponential data demand if hardware efficiency remains unchanged. Long-term energy balance will require both clean power deployment and radical changes in computing architecture, including neuromorphic chips and biological computing solutions.