Casimir Inc. did not announce a better battery. That would have been too ordinary, too trapped in the old humiliations of chemistry: degradation, replacement, charging cycles, mining, cost. Better batteries belong to the world as we already know it, where energy has to come from somewhere and the invoice always arrives.

On May 12, the Houston company announced a $12 million seed round to commercialize semiconductor chips that produce continuous electrical power from quantum vacuum fields. Its first product, MicroSparc, is described as a 5mm x 5mm chip designed to deliver 1.5 volts at 25 microamps, roughly 37.5 microwatts, enough for a narrow class of ultra-low-power electronics. The company says commercial availability is targeted for 2028. The pitch replaces the larger battery with a tiny square of silicon that behaves, in the company’s framing, like a solar panel for the dark.

If it works as described, this is not a gadget story but a tear in the bookkeeping. Industrial IoT sensors, implantable medical monitors, and other devices shaped by battery replacement would have to be reconsidered first. If it scaled beyond that, “battery” would begin to sound like a historical inconvenience.

Failure would not make the story meaningless. It would place Casimir inside a much older pattern of frontier physics, venture capital, climate anxiety, and the human desire to find a door in the wall of material limits.

The difficult part is that the claim lives in the narrow strip between “obviously absurd” and “not dismissible from a press release.” Modern technological belief does much of its business in that strip, which is crowded now.

The Casimir effect itself is real. Conductive boundaries placed extremely close together produce measurable forces from quantum field fluctuations. In standard quantum electrodynamics, though, the Casimir force is conservative. You do not get a continuous power source by cycling plates unless energy comes back into the system from somewhere. The problem arrives when the language shifts from force to power, and from power to product. Extracting continuous useful work from the vacuum, if that is indeed the source, would require a mechanism outside the ordinary Casimir accounting. Physics has room for strangeness, but the energy ledger still has to close.

Casimir’s public case has layers. Twelve million in seed funding led by Scout Ventures. A 2024 NSF SBIR Phase I award. Patents describing conductive walls, cavity gaps, antenna-like structures, and engineered geometry. A 2026 Physical Review Research paper from Harold “Sonny” White and coauthors modeling dynamic-vacuum emergent quantization.

That stack matters because it separates this from a basement perpetual-motion website with animated lightning bolts. The public evidence still has to answer one central question. Has an independent party measured continuous net electrical power from the device under conditions that rule out ordinary explanations?

Until that answer exists, the correct stance is audit rather than mockery. The audit has to be boring on purpose. Extraordinary energy claims usually fail in the measurement, not in the imagination. Picoamps can be discoveries or ghosts in the instrument. Millivolts can be physics or dirt. A beautiful theory can be a map of a country that no device actually inhabits. The most important test is also the simplest. Build the same setup with people who have no financial upside in the answer.

This is where the romance drains out of the room.

The more revealing question is why the announcement can circulate as a serious business story before that public audit is available. That is the real thesis. Part of the answer is scale. Casimir’s first target is not a power plant in a shoebox but tens of microwatts, the territory of small sensors and embedded devices. That restraint gives the story a kind of tactical credibility.

Nobody is promising to run Manhattan from a chip by Christmas. The product begins where energy is already strange, in unattended devices, buried sensors, and machines nobody wants to touch after deployment. Battery replacement is not glamorous, but it is one of civilization’s quietly stupid rituals. We scatter tiny machines through cars, factories, bodies, fields, warehouses, oceans, and infrastructure, then build maintenance regimes around feeding them, pretending the logistics are merely practical when they are also a confession about how much invisible labor our “wireless” world still requires.

A chip that never needs feeding is therefore not a fantasy of abundance at first. It is a fantasy of not sending a technician.

This is how exemption technologies enter the bloodstream, as technologies that promise to exempt us from a known constraint before becoming civilizational propositions. The ladder is always visible in outline, even when its base is measured in microamps.

Venture capital’s favorite story is small enough to prototype, large enough to mythologize. The addressable market begins with inconvenience and ends with civilization. The pitch does not need to prove the whole staircase today. It needs to make the staircase feel rational.

Climate pressure makes the staircase easier to see. Artificial intelligence is hungry for electricity. Battery supply chains depend on difficult minerals. Energy infrastructure has become an object of geopolitical anxiety. The promised clean, electrified, optimized future keeps arriving with lithium mines, transmission queues, transformer shortages, data-center load, and the occasional reminder that the weather is now part of the balance sheet.

Into that atmosphere, vacuum energy arrives as theology. Not in the church sense, but in the older one, a story about what the universe owes us.

The appeal is not only that the device might work. It might release us from the moral ugliness of ordinary power, from mines, pipelines, turbine blade graveyards, battery fires, grid upgrades, foreign suppliers, tradeoffs, and landscapes altered to keep the lights on. Energy without a visible wound. Power without politics. Extraction with no visible site and no accountable scar.

This is why the vacuum is such a perfect object of desire. It sounds empty enough to be innocent and deep enough to be infinite. It lets the culture imagine a source with no victims because it has no visible location. The claim moves through a public already trained to believe that the most important technologies look impossible until suddenly they do not. Rockets land on barges. Software predicts protein structures. Generative models write passable legal memos and bad wedding vows. The recent past has made skepticism feel, to some people, like a failure of imagination, which makes the confusion useful.

Yet imagination is abundant here. What is scarce is not ideas but discipline, the willingness to let a beautiful claim fail the measurement.

The tech economy has become very good at converting constraint into narrative. Batteries are heavy, so wireless freedom becomes the story. Data centers need power, so accelerated intelligence becomes the story. Climate systems are destabilizing, so clean abundance waits just over the next financing round. Each narrative may contain real engineering, which is what makes it powerful. The problem begins when the narrative starts doing work the engineering has not yet earned.

Casimir may have something. The NSF award has weight. White’s peer-reviewed paper has weight. The patent record has weight. The pile is insufficient, but it gives this claim a seriousness that pure perpetual-motion pitches lack. A published critical response to the paper does not appear to have surfaced yet. That absence says only that scrutiny is still young, not that the claim has survived it. That seriousness is what makes the story circulate, and what makes the eventual audit more interesting than a simple debunking.

The most interesting middle case sits between miracle and fraud: a device that produces a narrow, replicable anomaly for reasons the marketing language misnames. Maybe the source is an unrecognized conventional effect. Maybe it is fabrication chemistry, thermal behavior, environmental coupling, or some other mundane mechanism that still happens to be useful. In that world, the question shifts from “is this real?” to “what is this?” Those are very different futures. They should not be flattened into the same verdict.

What can be said now is modest and revealing. A slice of the market is ready for a company like this. It is ready to fund it, circulate the story, and treat the quantum vacuum as a commercial category.

Technological belief has always needed symbols. The steam engine offered controlled force. The microchip offered compressed intelligence. The battery became portable autonomy, a little brick of stored permission that lets the device leave the wall. A vacuum-energy chip, even at sensor scale, offers something new. Autonomy without storage. Power without replenishment. Infrastructure collapsed into matter itself.

It is a beautiful symbol. That is why it requires harsher evidence.

The danger is larger than belief in one strange energy claim. A culture under pressure can start preferring technologies that promise exemption from consequences over technologies that require discipline inside them. Grid upgrades are boring. Demand reduction is politically poisonous. Permitting reform is tedious. Efficiency is invisible when it works. Maintenance has no mythology. A chip that whispers power out of the vacuum has all the narrative force those things lack.

Maybe Casimir will earn that narrative. If it does, the first honest response should be astonishment followed immediately by replication. A national lab calorimetry test would matter. So would a blinded device comparison, a geometry-control experiment that kills the predicted effect, or a replication by a team with no financial stake in the outcome. The future has room for surprises. It has less room for evidence-free exemption, which is the one miracle no technology should get.

Until then, the vacuum bubble works as both physics story and cultural pressure gauge, a measure of how badly we want the universe to contain a loophole. We keep looking for an energy source clean enough to absolve us of appetite, small enough to fit inside the devices we refuse to stop multiplying, and strange enough to let us believe the old accounting has been suspended.

The chip may or may not produce power. The belief already does.

The Research Trail

Read more

Researchers gave bacteria a job, watched them do it, and watched them leave over time when dosing stopped. The job was happening inside a living gut. The exit was part of the therapeutic claim. In a Cell paper published in April 2026, scientists modified strains of Lactobacillus plantarum for hepatic encephalopathy, a condition where a failing liver turns into a failing mind. The bacteria did more than deliver a drug. They performed metabolic labor in place, handling ammonia, shifting amino acid balance, changing what a failing mind could still do in mouse models. Medicine as municipal service, a sanitation department with plasmids.

The study remains preclinical. Mouse-model success is exactly where the fantasy of control gets overconfident. A pill acts as an event. It enters, dissolves, binds, blocks, stimulates, then leaves behind improvement, side effects, or a bill. These bacteria behaved less like an event than a temporary metabolic tenant. They lived, did work, and, crucially, cleared gradually after dosing stopped. That last detail carries the plot, because exit is not a side note in living medicine; it is the product claim.

The early fear of living medicine is gothic. Swallow a creature, let it colonize you, hope the creature remains polite. The Cell paper points toward something stranger than permanent occupation, a medicine that lives for a while, responds inside the body, and turns treatment into a relationship among organism, host, microbiome, diet, immune state, and time. When the boundary around “the drug” starts to leak, institutions reach for rituals that make a thing fileable. They want a name, a protocol, a signature, and a box to put it in.

Boundaries make medicine governable. Even biologics mostly preserved the grammar of the dose. Make a thing, characterize it, administer a quantity, measure effects. When the effects turn bad enough, stop the thing, clear it, block it, or wait for it to decay. The patient may not feel in control, but the regulator still has a handle.

Engineered living therapeutics loosen that handle. They are compounds with agency, able to respond to local conditions and, in some designs, replicate while doing the work. The promise is precision so local it starts to look like citizenship. The microbe lives where the problem is, reads the local chemistry, works only there. That is the point where medicine starts needing house rules for a tenant with metabolism.

The promise is scientific. It’s also political. Before anyone signs off on a living intervention, they need a fiction of control sturdy enough to sign. The microbe must look less like a guest and more like a contractor with a badge, a scope of work, an exit interview, and a very clear answer to what happens if it leaves the job site.

The FDA is not starting from zero. Its 2016 guidance on live biotherapeutic products tells sponsors what early trials need to show about manufacturing and control. The agency has approved microbiota products like Rebyota and Vowst, fecal-derived communities rather than engineered single-strain organisms. The paperwork exists. That actually matters. So do the definitions, committees, and spreadsheets with tab names like “fecal product comments final FINAL.”

A category tells the agency where to file the thing. For the patient, hosting it opens the gap where trouble starts. Agencies approve products, clinicians prescribe courses, insurers reimburse codes, and patients are asked to make room inside their bodies for the thing everyone else has named.

Most regulatory systems work best with products you can specify before use. Living therapeutics keep producing relevant facts after they have been swallowed. A 2024 EMBO Molecular Medicine review names the anxiety. Bacteria, phages, and their genes can multiply, evolve, and spread. Safeguards exist. For now. Some strains can be made dependent on molecules missing outside the target setting. Kill switches can make escape trigger death. Mutations can disable a safeguard, and gene transfer can sometimes undo containment assumptions. Containment is a treaty with evolution, renegotiated every generation.

Conventional drugs have their own talent for mayhem. A small molecule that wrecks a liver is not morally superior because it lacks a genome. But living therapeutics change the failure mode. A chemical side effect asks what damage the substance caused. A living side effect asks what the system became.

The target diseases are where ordinary medicine already fails. Hepatic encephalopathy pulls the gut, liver, and brain into the same breakdown. The person becomes a cascade of failing organs. If engineered commensals can reduce ammonia while preserving microbiome diversity, even in animals, the appeal is obvious. Instead of carpet-bombing microbial communities with antibiotics or forcing one liver pathway harder, redesign part of the metabolic neighborhood.

The gut keeps attracting this metabolic ambition because digestion is only the obvious part. The gut negotiates the outside world into the body. Border control, waste management, immune training, neighborhood politics conducted in mucus. Medicine has spent a century treating the body as a set of organs. The microbiome makes that map bureaucratically useful and biologically incomplete.

There have already been warnings. Synlogic’s SYNB1020, an engineered E. coli Nissle designed to consume ammonia, produced early safety data in healthy volunteers, then failed to lower blood ammonia in a Phase 1b/2a trial in patients with cirrhosis. The company stopped the program in 2019. That does not make the field a dead end, because Synlogic later advanced SYNB1934 for phenylketonuria into a pivotal Phase 3 study, targeting a different metabolite in a different disease, but it does keep the SYNB1020 failure at the center of this story as the cleanest caution against treating gut chemistry as a clean bioreactor. The patient’s illness is part of the machine the organism must negotiate.

“Programmability” sounds like software. It should sound more like diplomacy. Instructions can be given to the organism, but the host cannot be repealed. The person hosting the system still needs a way to stop it.

Consent works best when the intervention has a perimeter. Surgery has a wound and a scar. A pill has a dose and a half-life. A device has a model number and a replacement schedule. Microbial medicine has a different shape. The patient is asked to sign an agreement among parties they cannot see, host, strain, microbiome, clinician, regulator, insurer, time. “Informed consent” starts to sound less like knowledge than permission to enter a room where the lights are off.

This doesn’t make consent impossible. It makes reversibility central. The most important product feature in living medicine may not be what the microbe can do but whether the patient can stop being its habitat. Clearance after dosing, genetic safeguards, antibiotic sensitivity, long-term monitoring. These are not boring technical details. They keep “yes” attached to the person who said it.

The class politics are not proven by the science. They are the incentive path to watch if living medicine becomes ordinary enough to bill. The first versions of programmable bacteria will treat serious disease. Later versions may tempt medicine toward murkier categories. The same logic that lets a microbe consume ammonia could tune inflammation, metabolize dietary compounds, release neuroactive molecules, alter satiety signals, modulate immune tone. Some goals will count as treatment. Others will call themselves prevention, wellness, or enhancement with a reimbursement code. Once biology becomes infrastructure, infrastructure acquires service tiers.

The wealthy already buy cleaner air, quieter neighborhoods, concierge medicine, embryo selection, continuous glucose monitors, and sleep optimization. If programmable microbiomes work well enough to become status markers, they will fit neatly into that economy. The rich will not call it installing custom bacteria. They will call it maintaining metabolic resilience. The rest of us may call our insurer and learn that our gut flora is out of network.

“Medical necessity” could become the polite phrase for biological inequality. The same microbial circuit might look like care when a patient is sick, prevention when an affluent patient wants cover, optimization when cash changes hands. The organism performs the same chemistry in all three bodies. The billing code does the moral sorting before anyone has to admit what is being sorted.

Regulators will try to draw lines between therapy and enhancement, transient and persistent, natural and engineered, drug and food. Reality will walk through them. A microbe does not care whether its metabolic output has a treatment, prevention, or optimization label.

The best rulebook will treat living therapeutics as biological relationships, not ordinary drugs with a pulse. It will ask where the organism goes, how it stops, when patients are notified, and who pays when it misbehaves.

The Cell study does not announce the age of permanent bacterial colonization. It announces the opposite: a future in which engineered microbes must prove they can leave. Regulators need measurable exits. Companies need clearance endpoints they can print on the label. Clinicians need an explanation they can give without crossing their fingers. Patients need the exit to work in the body, not just in the protocol.

The living pharmacy will not look like a monster movie. It will look like a capsule, consent paperwork, a tracking portal, and a clinician explaining that the organism has been designed to clear after the course ends. Then everyone in the room will pretend that “designed to” is the same as “will.”

The better question arrived once the sources started pushing back. When medicine uses gut bacteria as programmable metabolic tools, how do we decide when a living system is controllable enough?

That question became useful because it forced every source to answer the same practical problem from a different angle, moving from molecular design to animal evidence, then into regulation, containment, clinical failure, and consent.

The useful sources were not the ones that made engineered microbes sound most futuristic, but the ones that kept changing the scale of the problem, from a strain doing chemistry in a mouse gut to an agency deciding what kind of evidence can make a living product feel bounded enough for people to accept.

The Research Trail

Read more

Longevity is the polite name for an old wish, more life, but not just more years at the frail end. The version now attracting money and attention is about extending healthy years. Less dementia. Less cancer. More time before the body becomes the main fact of a life.

There are good reasons to talk about it now. Aging biology has become less mystical. Scie…

Read more

The Research Trail

Read more

The Sahara is the wrong kind of sand. So is the Arabian Peninsula. The Gobi. The Atacama, and every other desert on earth. These places together represent a third of Earth’s land surface, blanketed in the material modern civilization needs most, and almost none of it is usable. Desert sand is round. Billions of years of wind erosion smooth each grain in…

Read more

Read more

On the evening of April 10, a capsule named Integrity hit the Pacific. Four people climbed out onto the deck of the USS John P. Murtha: Reid Wiseman, Victor Glover, Christina Koch, Jeremy Hansen. They were the first people to travel beyond low Earth orbit in fifty-four years. They had covered roughly 695,000 miles, pushing the farthest-from-Earth record…

Read more

The research trail

First stop, the NASA Artemis II launch day blog and the splashdown-day news release. Crew (Wiseman, Glover, Koch, Hansen), closest far-side approach 4,067 miles, distance record set April 6.

Read more

In January, researchers at Carnegie Mellon and Fujitsu presented FieldWorkArena at the AAAI conference in Singapore. A benchmark designed to measure whether AI agents are safe enough to field in live industrial settings. Factories. Warehouses. Places where the wrong answer doesn’t embarrass the product manager but puts someone in a sling.

FieldWorkArena …

Read more

That question opened up a lot of territory. Here’s how the trail ran.

The Research Trail

Read more

Kelvin signs the water permit on a Tuesday. It’s a good permit. The engineering is sound, the flow rates are within spec, the aquifer recharge modelling says the draw is sustainable at the facility level. He’s a competent man doing competent work. He has a mortgage in suburban Virginia, and the permit crosses his desk alongside eleven others that week, …

Read more

Read more

On January 30, 2026, SpaceX filed with the Federal Communications Commission for permission to launch up to one million satellites into low Earth orbit. Not for internet access. For data centers. The FCC accepted the application five days later, classifying the deployment under a categorical exclusion from environmental review. The reasoning: activities…

Read more

Read more

Every Atlassian will receive an email within the next twenty minutes.

That sentence dropped in a pre-recorded video from CEO Mike Cannon-Brookes on March 11, 2026, moments before 1,600 people learned they no longer worked at a collaboration software company. Twenty minutes. Enough time to make coffee. Not enough time to find out who you are without the b…

Read more

Read more

Bastiat’s job title didn’t exist eighteen months ago. Now there are LinkedIn influencers posting about it.

What she does, in practice, is translate. Not languages. Tiers. When the property management platform’s pricing agent reaches tacit equilibrium with every other agent in a metropolitan area, without exchanging a single message, Bastiat writes the co…

Read more

Not a hypothetical. Visa is already building agent commerce infrastructure across an ecosystem of over a hundred partners, with payment rails designed for agent-to-agent transactions. PYMNTS reported that commerce is shifting from clicks to prompts, with…

Read more