Powering the machines inside the human body
Every machine sealed in the human body lives on a starvation energy budget. This essay follows the watt: why batteries force repeat surgery, how power crosses living tissue, why plutonium once solved the problem, and which technologies are trying to retire the battery in 2026.
Most pacemaker surgeries are not repairs. The device in the patient’s chest works exactly as designed. It senses every beat, paces on schedule, logs its own telemetry. It goes back under the knife for one reason: the battery is running down, and the battery is sealed inside the can with everything else. Every seven to ten years or so, the whole unit comes out and a new one goes in, with the infection risk, the cost, and the anesthesia that any surgery carries. Multiply that across the enormous fleet of cardiac devices in service, add the neurostimulators, the cochlear implants, the drug pumps and the experimental brain interfaces, and a strange picture emerges. The binding constraint on machines that live inside the body is not computation, not sensing, not electrode chemistry. It is the watt.
This is worth taking seriously, because the power problem shapes everything else. It decides how big an implant is, since the battery usually dominates the volume. It decides how smart an implant can be, since every instruction executed is energy spent. It decides how much data a neural interface can send out of the skull, since radio is the hungriest thing an implant does. And it decides how often a human being goes back to the operating room to service a machine that never broke.
A starvation budget
The numbers involved are almost comically small. A modern pacemaker runs on a continuous budget in the neighborhood of ten microwatts, a millionth of what a phone charger delivers. It achieves this by doing almost nothing almost all the time: a brief sense, a decision, and only when needed a pulse of a few volts into high impedance tissue. Whole engineering careers have gone into shaving nanoamps off its sleep current.
Climb the ladder and the budget grows fast. A spinal cord or deep brain stimulator that fires continuously sits in the milliwatt range, which is why many of them ship with rechargeable cells and an induction paddle the patient wears against the skin. A cochlear implant processes audio in real time and stimulates dozens of electrode sites; it solves the problem by refusing to internalize it, keeping the battery in the external sound processor and beaming power through the skin continuously. And a high channel count brain interface that digitizes, compresses and transmits neural data pushes into tens of milliwatts, which collides with a hard biological ceiling: dissipate too much heat in living tissue, on the order of a degree or two of warming, and you begin to damage the thing you are recording from. The power budget is not just about supply. It is a thermal allowance, and the brain grants a very small one.
The ladder in the figure explains most of the field’s structure at a glance. A source can only serve loads at or below its own rung. Betavoltaic cells and biofuel cells live near the bottom, so they can pace a heart but not stream a cortex. Wireless links reach the top rungs, but only while an external transmitter is present and aligned. Every practical implant is a negotiation between these columns.
Getting watts through meat
If the battery is the problem, the obvious move is to leave the energy outside and send it in. The physics of that transfer is unforgiving, and it is worth seeing why.
Any energy beam crossing tissue is absorbed along the way, and the absorption is exponential: the intensity that survives a depth d goes as I = I0 e^(-a d), where a is the attenuation coefficient of the tissue at that frequency. In plain terms, every additional centimeter of muscle takes the same fixed percentage off the top, so losses compound the way debt does, and doubling the depth squares the surviving fraction. For radio frequencies high enough to focus on a small receiver, a in tissue is brutal, and most of what you transmit becomes heat in the flesh above the implant, which is exactly where you do not want it.
Magnetic induction, the technology in cochlear implants and rechargeable stimulators, sidesteps absorption by working in the near field at low frequency, where tissue is nearly transparent to magnetic flux. The price is geometry: coupling between two coils collapses rapidly with distance and misalignment, so induction works beautifully through a centimeter of skin with a magnetically aligned puck, and poorly for anything deep or mobile.
Ultrasound is the interesting middle path. Sound at megahertz frequencies propagates through soft tissue with far less absorption than microwaves, and its millimeter wavelengths focus onto receivers the size of a grain of rice. A piezoelectric crystal at the focus turns pressure back into electricity. The approach has produced the smallest powered implants ever demonstrated, motes deep in tissue with no battery at all. Its weaknesses are equally physical: ultrasound does not cross air or bone gracefully, the transmitter must know where the receiver is, and the duty cycle is hostage to whatever holds the transducer against the skin.
The plutonium precedent
The battery problem is old enough to have produced one genuinely radical solution, and it is worth remembering how far medicine was once willing to go. Beginning in 1970, several thousand patients received pacemakers powered by plutonium-238, a few curies of it, whose steady radioactive decay heat was converted to electricity by thermocouples. The isotope’s half life is 88 years, so the power source outlived not only the device but usually the patient, as documented in the Oak Ridge health physics museum’s collection. One patient was paced by the same nuclear unit for 35 years, reported in the cardiology literature as the longest run of its kind. These devices carried federal tracking obligations to the grave and beyond: on the patient’s death the generator was supposed to be recovered and shipped back for the plutonium to be reclaimed, a logistics tail described in the American Nuclear Society’s account of one such recovery.
What killed the nuclear pacemaker was not radiation. Shielded, the dose to the patient was small, and the safety record was good. It lost to the lithium iodine battery in the late 1980s for reasons that had nothing to do with physics: licensing, tracking, customs paperwork every time a patient crossed a border, and the sheer institutional weight of owning plutonium in thousands of ambulatory chests. The lesson generalizes. In implanted energy, the constraint that decides is rarely the elegant one.
Where this stands in 2026
Three currents define the moment, and each one attacks a different rung of the ladder.
Betavoltaics are back. These are not thermal generators like the plutonium units but semiconductor devices that harvest electrons directly from beta decay, typically nickel-63 layered onto a diamond or silicon junction. Chinese firm Betavolt has been promoting its coin sized BV100, claiming decades of maintenance free output at microwatt scale, figures that remain the vendor’s own, as covered by Live Science. In February 2026, a US startup operating as Project Omega came out of stealth with a nickel-63 betavoltaic line delivering nanowatts for a claimed century of life, alongside an isotope recycling business, reported by Chemical and Engineering News. The same reporting puts nickel-63 near four thousand dollars a gram, which frames the real question: not whether decay power works, but what a microwatt is worth. At today’s numbers, a betavoltaic source sits one comfortable rung above a pacemaker’s needs on paper and one uncomfortable rung below its pulse currents, so the near term designs pair the isotope with a small buffer cell, decay trickling into a capacitor the way a spring fills a cistern.
Ultrasound power is maturing from lab trick to engineering discipline. Work published this year on flexible acoustic metamaterials shows lightweight lenses that refocus ultrasound through tissue onto an implanted receiver, recovering much of the energy that plane wave transmission wastes. Korean groups have demonstrated layered piezoelectric harvesters that, per the research institute’s announcement, can recharge a commercial implant battery through tissue in around two hours, a lab figure but a striking one. The architecture this enables is a hybrid: a small internal battery for autonomy, topped up wirelessly on a schedule, with no surgical replacement in the device’s lifetime.
The third current is the strangest: making the body the power plant. Glucose biofuel cells oxidize the sugar already circulating in extracellular fluid, using enzyme or nanomaterial electrodes. The landmark animal result remains a cell implanted in a rat that produced tens of microwatts continuously and ran an LED, published in Scientific Reports, and a comprehensive 2024 review in Heliyon maps a decade of progress since. A working glucose cell is thermodynamically beautiful: the implant eats what the patient eats, forever. It is also the least ready of the three, because enzymes denature, electrodes foul, and the immune system treats every surface as a construction site.
What breaks
Each of these stories carries its own asterisk, and they should be said plainly. Betavoltaic power densities are orders of magnitude below marketing intuition; no isotope cell demonstrated to date streams data or drives continuous stimulation on its own, and the regulatory pathway for new radioactive implants, post plutonium, is uncharted rather than merely slow. Ultrasound charging figures almost always come from well aligned laboratory geometries with the transducer clamped in place; a patient leaning in a chair is a different acoustic channel. Glucose cell lifetimes in living animals are still measured in months, not decades. And the field’s history argues for humility: the plutonium pacemaker was a technical success retired by paperwork, a reminder that the best energy source is the one an insurer, a regulator and a customs officer can all live with.
Still, the direction is unmistakable. The battery earned its place in the body by being the least bad option in 1985, and for forty years the surgical replacement cycle has been treated as a law of nature. It is not. It is a consequence of sealing the energy inside the can, and every one of the technologies above is a different way of unsealing it. The implants of the next decade will be judged, as much as anything, by how rarely they force a healthy person back onto the table.