News / Updates / Blog:

  • Anti-Poaching Dogs in Africa: How Trained K-9s Are Protecting Endangered Species

    One anti-poaching dog, in favorable conditions, can secure a wildlife habitat of up to 32 square kilometers with the search capability of roughly 60 human rangers covering the same ground over the same period. The dog can track a scent trail that’s 20 to 40 hours old. It can chase a target at 32 kilometers per hour. It exerts 240 pounds per square inch of bite pressure. It cannot be corrupted, bribed, or intimidated, and it will work seven days a week provided it gets eight hours of rest and adequate care. Since K-9 units were introduced to South Africa’s national parks in 2012, dogs have been involved in 80 percent of poacher apprehensions in the areas where they operate.

    Those numbers matter because the thing they’re protecting against is not abstract. Rhino horn sells for approximately $65,000 per kilogram on the black market—more expensive per gram than gold or cocaine. A single horn weighs six to seven kilograms. At the start of the 20th century, roughly 500,000 rhinos roamed Africa and Asia. By 1970, that number had dropped to 70,000. Today, approximately 27,000 remain on the entire planet, and South Africa—home to about 80 percent of the world’s rhinos—has been the epicenter of a poaching crisis that exploded in 2010 and hasn’t stopped. The dogs didn’t solve the crisis. But they changed the math in the places where they operate, and the way they changed it tells you something about why animals remain indispensable tools in contexts where technology alone can’t do the job.

    Why dogs and not drones

    Kruger National Park is roughly the size of Israel. It’s largely wilderness—thick vegetation, rugged terrain, limited road infrastructure—and poachers enter on foot, often at night, moving through bush that provides near-total concealment. Aerial surveillance can cover large areas but can’t penetrate tree canopy. Thermal imaging helps at night but generates false positives from every warm-blooded animal in the park. Ground sensors detect movement but can’t distinguish a poacher from a warthog. GPS tracking requires something to track—it’s useless against people who aren’t carrying devices.

    A dog’s olfactory system processes scent with roughly 300 million receptor cells, compared to about 6 million in humans. The portion of a dog’s brain devoted to analyzing scent is proportionally 40 times larger than a human’s. This isn’t a marginal advantage. It’s a different category of sensory capability, and in an environment where visual detection is limited by vegetation and darkness, scent is the primary information channel. A poacher who entered the park eight hours ago, walked five kilometers through the bush, and is now lying still in a thicket is invisible to cameras, drones, and rangers with binoculars. He is not invisible to a dog.

    The breeds used across African anti-poaching operations are selected for specific roles. Belgian Malinois dominate—they’re fast, driven, aggressive when needed, and bond intensely with their handlers. Doberman-bloodhound crosses are used as cold-spoor trackers, combining the bloodhound’s extraordinary olfactory organ with the Doberman’s lean build and high drive. The South African organization Pit-Track breeds these crosses specifically for anti-poaching work and has distributed 33 dogs to units across five African countries. Labradors and spaniels work as detection dogs, sniffing out rhino horn, elephant ivory, pangolin scales, firearms, and ammunition during vehicle searches and at transit points. Each breed fills a different operational niche, and the units that perform best use them in combination—trackers to follow the trail, patrol dogs to apprehend, detection dogs to find concealed contraband.

    How the operations actually work

    A typical anti-poaching response begins when rangers discover evidence of incursion—fresh footprints, cut fences, or a poached animal. The K-9 team deploys to the entry point. The tracking dog picks up the scent trail and follows it, often for kilometers, through terrain that would take human trackers hours longer to cover. If the poachers are still in the park, the dog closes the distance faster than they can move on foot. If they’ve already exited, the dog can track them to their exit point, often providing enough evidence—direction of travel, vehicle tracks, dropped items—to support investigation and arrest.

    The “high-speed” tracking dogs developed at the Southern African Wildlife College have been described by trainers as the single biggest game-changer in the counter-poaching toolkit. These dogs are released off-leash and trained to pursue and hold a target—barking to alert handlers, or physically engaging if the poacher runs or fights. The psychological effect on poachers has been significant. Multiple reports from ranger units describe poachers altering their tactics specifically to avoid areas known to have K-9 units, preferring to operate in parks or conservancies without dogs. The deterrent value may be as important as the apprehension value.

    The Anti-Poaching Tracking Specialists in Zimbabwe’s Savé Valley Conservancy—one of the largest private game reserves in Africa at over 1,150 square miles—have used an 11-dog Belgian Malinois unit to crack dozens of rhino poaching syndicates. Their operations resulted in 29 rhino poacher arrests in four years, contributing to a cumulative 189 years of prison sentences. Their lead handler, Mathius Mbengo, and his K-9 partner cover roughly 15 kilometers per day on foot, tracking through bush terrain. At Ol Pejeta Conservancy in Kenya, only one rhino has been poached in two years since the introduction of dogs. At Mkomazi in Tanzania, there have been no poaching incidents in the seven months since dogs caught the last bushmeat poaching gang.

    The organizations and the scale

    The K-9 anti-poaching ecosystem in Africa is a patchwork of government agencies, NGOs, private conservancies, and international donors, each running their own programs with varying levels of funding, training quality, and operational integration.

    SANParks—South Africa’s national parks authority—established its K-9 unit in 2012 with a handful of dogs in Kruger. By 2016, roughly 60 dogs were working across the park. The SANParks Honorary Rangers’ K9 Project Watchdog, a volunteer-supported initiative, now operates across eight national parks—seven rhino parks and Table Mountain National Park, where the target species is abalone, a marine mollusk heavily poached for East Asian markets. The project purchases dogs, builds kennels, covers veterinary costs, and provides equipment and training for handlers.

    The Black Mamba Anti-Poaching Unit, founded in 2013 and operating in the Balule Nature Reserve and Greater Kruger area, is notable as a predominantly female ranger force—a detail that challenges assumptions about who does this work and how. Animals Saving Animals, founded in 2016, has placed dogs in anti-poaching operations from South Africa to Costa Rica. Dogs4Wildlife operates in Tanzania, Zimbabwe, South Africa, and Rwanda. The Sheldrick Wildlife Trust runs tracker dogs alongside its elephant orphan program in Kenya, using them to find ivory, rhino horn, bushmeat, and firearms.

    The limitations

    Dogs aren’t a solution. They’re a force multiplier within a solution that requires funding, governance, community engagement, demand reduction, and law enforcement capacity that most affected countries struggle to maintain. A dog costs roughly $25,000 to purchase and train, plus ongoing veterinary care, handler salary, and operational support. That’s cheap relative to a helicopter but expensive relative to what most African parks can afford without external donor funding. Dogs need rest, veterinary attention, and handlers who are trained, motivated, and not themselves vulnerable to corruption—a real concern in regions where a single rhino horn is worth more than a ranger’s annual salary.

    The poaching networks are transnational criminal enterprises with supply chains stretching from bush trackers in Mozambique to horn dealers in Vietnam and China. Catching the person with the machete in the park addresses the immediate threat but doesn’t touch the demand signal or the intermediary networks that move product across borders. Dogs are a tactical asset. The strategic problem—a global market that assigns a per-kilogram value to rhino horn exceeding the per-kilogram value of cocaine—requires economic, diplomatic, and law enforcement interventions that no animal can provide.

    But in the space between the poacher’s entry into the park and the moment they reach the rhino, a Belgian Malinois running flat-out through the bush at midnight is the most effective intervention that currently exists. The technology is four legs, 300 million olfactory receptors, and a relationship between a dog and a handler built on thousands of hours of training and mutual trust. It’s not scalable the way a sensor network is scalable. It’s not deployable the way a drone fleet is deployable. It’s effective in the way that a living organism with millions of years of evolutionary optimization for exactly this kind of work is effective—which is to say, in ways that engineered systems can’t yet replicate and may never fully replace.

    We cover anti-poaching K-9 units alongside military dolphins, landmine-detecting rats, and a dozen other cases of animals deployed in service of human objectives across our Animal Heroes course—including why the most sophisticated sensor platform in the African bush weighs 30 kilograms and answers to the name Bandit.

  • Hypersonic Weapons in 2026: Who Has Them and Do They Actually Work?

    The United States has spent over $15 billion on hypersonic weapons development across the last three fiscal years. The Pentagon‘s FY2026 budget requests another $3.9 billion. And as of March 2026, the United States has not fielded a single operational hypersonic weapon. Russia claims to have several. China has tested at least 20 times more hypersonic missiles than the U.S. has. A former Under Secretary of Defense for Research and Engineering testified to Congress that the United States does not “have systems which can hold [China and Russia] at risk in a corresponding manner, and we don’t have defenses against [their] systems.”

    That quote is from 2023. The situation in 2026 is marginally better and structurally the same.

    What hypersonic actually means

    The term gets thrown around loosely enough that it’s worth nailing down. A hypersonic weapon travels at Mach 5 or above—five times the speed of sound, roughly 6,200 kilometers per hour at sea level. But speed alone isn’t the point. Conventional ballistic missiles reach hypersonic velocities during reentry. What distinguishes hypersonic weapons is that they combine extreme speed with maneuverability—the ability to change course during flight, making their trajectory unpredictable and interception extraordinarily difficult.

    Two types dominate the current development landscape. Hypersonic glide vehicles are launched on a ballistic trajectory by a rocket booster, then separate and glide through the upper atmosphere at hypersonic speeds, maneuvering toward their target along a flatter, lower path than a traditional ballistic missile. This depressed trajectory reduces the detection window for radar systems and complicates interception because the glide vehicle can change direction in ways a ballistic warhead can’t. Hypersonic cruise missiles are powered throughout their flight by air-breathing engines—scramjets—that sustain hypersonic speeds within the atmosphere. They fly lower than glide vehicles, further reducing detection time, but the engineering challenge of a scramjet engine that operates reliably at Mach 5-plus is formidable.

    The strategic concern is straightforward: a weapon that moves at Mach 5 or faster while maneuvering unpredictably compresses decision-making time for the target country to minutes or less, and existing missile defense systems—designed to intercept projectiles on predictable ballistic trajectories—can’t reliably track or engage it.

    Russia: Claims and combat use

    Russia has been the most aggressive in deploying claimed hypersonic capability. The Kinzhal (meaning “dagger”), an air-launched ballistic missile reportedly capable of Mach 10, has been used operationally in Ukraine since March 2022. The Avangard hypersonic glide vehicle, which Russia claims can reach Mach 27 and maneuver to evade any missile defense system, was declared operational in December 2019, deployed with the Strategic Rocket Forces on modified SS-19 Stiletto ICBMs. In August 2025, President Putin announced that the Oreshnik hypersonic ballistic missile had entered production and would be deployed in Belarus.

    The caveat on Russian claims is that independent verification is limited. Russia has a history of announcing capabilities in advance of demonstrated performance, and the operational record of Russian advanced weapons in Ukraine has been mixed. Kinzhal has been intercepted by Ukraine’s Patriot missile defense systems on at least two confirmed occasions, which either means the Patriot is more capable against hypersonic targets than expected or the Kinzhal is less capable than advertised—or both. The Avangard has been tested only twice, with one success and one failure. Production quantities for all Russian hypersonic systems are believed to be very small, constrained by sanctions, manufacturing capacity, and component shortages.

    China: The scale advantage

    China has conducted the most extensive hypersonic testing program of any country—up to 20 times as many tests as the United States, according to Congressional Research Service reporting. In late September 2025, China conducted a hypersonic ICBM test featuring boost-glide technology and a depressed trajectory, which analysts interpreted as a major leap in flight-profile sophistication.

    China has also unveiled multiple operational or near-operational hypersonic systems: the YJ-17, an anti-ship aeroballistic missile with a hypersonic glide vehicle warhead; the YJ-19, a scramjet-powered hypersonic cruise missile; and the CJ-1000, a hypersonic cruise missile designed to target systems nodes across land, sea, and air. China’s DF-17 medium-range ballistic missile, which carries a hypersonic glide vehicle, has been deployed since 2020 and is specifically designed to threaten carrier strike groups and military installations across the western Pacific.

    The difference between the Chinese and American approaches is structural, not just technical. Chinese and Russian hypersonic weapons are designed for nuclear or dual-capable use, which means they can afford to be less accurate—a nuclear warhead doesn’t need to hit a specific building. American hypersonic weapons are being designed exclusively for conventional warheads, which means they require far greater precision. As the CRS report notes, nuclear-armed systems can be 10 to 100 times less accurate than their conventional equivalents. The U.S. has chosen the harder engineering problem.

    The United States: Expensive, capable, and not yet fielded

    The U.S. has three primary hypersonic weapons programs in development. The Navy’s Conventional Prompt Strike pairs a common hypersonic glide body with a booster system, intended for deployment on Zumwalt-class destroyers (now delayed to 2027) and eventually Virginia-class submarines and Burke-class destroyers. The Army’s Long-Range Hypersonic Weapon—Dark Eagle—is a road-mobile, trailer-launched system with an estimated range of 3,500 kilometers. In December 2025, Lt. Gen. Francisco Lozano disclosed that Dark Eagle could strike mainland China from Guam, Moscow from the United Kingdom, or Tehran from Qatar. One battery is stationed at Fort Lewis, Washington. The Air Force’s Hypersonic Attack Cruise Missile, a scramjet-powered system, is the service’s primary focus after the cancellation of the ARRW program following multiple test failures.

    The U.S. conducted successful end-to-end tests of the common glide body in June and December 2024 and April 2025—genuine milestones after the embarrassing 2022 test failure and the 2023 cancellation due to a battery issue. But operational deployment remains years away, and CRS analysis states the U.S. is unlikely to field an operational hypersonic system before FY2027.

    The cost problem may be more consequential than the timeline problem. Current U.S. hypersonic missiles cost between $15 million and $41 million per round, depending on the system. Production capacity is one to two missiles per month. An Atlantic Council task force, co-chaired by former Air Force Secretary Deborah Lee James and former Army Secretary Ryan McCarthy, published a report in late 2025 arguing that the U.S. defense industrial base is structurally incapable of producing hypersonic weapons affordably or at scale. Former Pentagon hypersonics director Michael White said the U.S. has “very capable” systems but at “unsustainable cost levels” and urged a “dramatic shift” toward commercial manufacturing models. In February 2026, Ursa Major debuted the HAVOC missile system—a medium-range hypersonic weapon explicitly designed for affordability and mass production, including the ability to be launched from fighter aircraft, ground systems, and space.

    The defense problem

    Building a weapon that existing defenses can’t stop is one challenge. Building a defense against that weapon is arguably harder. The Pentagon is conceptualizing a multi-layered intercept architecture called “Golden Dome,” combining space-based sensors, midcourse interceptors, high-altitude systems, and terminal-phase defenses. The U.S. has procured upgraded AN/TPY-2 radars with gallium nitride arrays for hypersonic threat detection. The Missile Defense Agency requested $200.6 million for hypersonic defense in FY2026.

    The fundamental physics problem: intercepting a maneuvering object at Mach 5-plus requires the interceptor to be faster and more agile than the target, or to predict its trajectory accurately enough to position the interceptor in its path. Hypersonic glide vehicles are specifically designed to make trajectory prediction impossible. The interception window is measured in seconds. The sensor coverage required to track a low-flying, maneuvering target across thousands of kilometers doesn’t exist in any deployed system. Patriot’s successful intercepts of Kinzhal suggest the problem isn’t completely unsolvable, but Kinzhal may be the least sophisticated of the deployed hypersonic threats.

    The arms control vacuum

    The New START Treaty between the U.S. and Russia, which limited strategic nuclear weapons, expired in February 2026 without renewal. The treaty didn’t explicitly cover hypersonic glide vehicles or cruise missiles that fly on a non-ballistic trajectory for more than half their flight. No international agreement restricts the development, testing, or deployment of hypersonic weapons. North Korea, India (developing BrahMos II with Russia, targeting Mach 7), Japan (the HVGP, expected in service by 2026), France, Australia, and Iran are all pursuing hypersonic programs at various stages of maturity.

    The proliferation trajectory mirrors what happened with ballistic missiles in the mid-20th century and drones in the 2010s: a technology initially limited to a few major powers diffuses to regional actors within a decade, and the window for arms control closes before anyone seriously tries to open it. With no treaty framework, no verification mechanisms, and no diplomatic momentum toward either, the hypersonic arms race in 2026 is entirely unregulated and accelerating.

    We cover hypersonic weapons alongside directed energy, autonomous systems, electronic warfare, and the full landscape of emerging military technology across 24 lectures in our Battlefields of the Future course—including why the most expensive weapons in the U.S. arsenal might be the ones it can least afford to use.

  • Solid-State Batteries in 2026: Why They’re Taking So Long and What Changes When They Arrive

    Toyota first announced it would have solid-state batteries in production by 2020. That was pushed to 2023. Then 2026. On October 7, 2025, Toyota’s solid-state battery officially received production approval in Japan—a genuine milestone, and one that arrives roughly five years behind the original schedule. Small-scale production is now confirmed for 2026–2027, with mass production planned for 2027–2028, in partnership with electrolyte supplier Idemitsu Kosan and cathode-material supplier Sumitomo Metal Mining. The target specs: 450 to 500 watt-hours per kilogram energy density, 10-minute charging to 80 percent, 1,000-kilometer range, a lifespan that Toyota’s chief battery engineer described as “maybe 40 years at 90 percent capacity.”

    If those numbers hold—and that’s a significant “if” given the history of solid-state battery announcements—they represent roughly double the energy density of current lithium-ion cells, five times the charging speed, more than double the range, and a battery that would outlast the car, the car that replaces it, and possibly the car that replaces that one. The promise is genuinely transformative. The reason it’s taken this long is genuinely difficult.

    The one-sentence version of the problem

    A solid-state battery replaces the liquid electrolyte in a conventional lithium-ion cell with a solid material. That’s it. That’s the entire conceptual leap. Everything else—the higher energy density, the faster charging, the improved safety, the longer lifespan—follows from that single substitution. And the reason it’s taken decades to commercialize is that making solid materials behave like liquids at the atomic level, inside a battery, under repeated charge-discharge cycling, at automotive scale and cost, turns out to be one of the harder materials science problems of the 21st century.

    In a conventional lithium-ion battery, lithium ions move through a liquid electrolyte between the anode and cathode during charging and discharging. The liquid electrolyte is flammable, which is why lithium-ion batteries occasionally catch fire—thermal runaway, in the technical term. The liquid also limits the anode material to graphite, because lithium metal anodes (which would dramatically increase energy density) form dendrites—metallic whiskers that grow through the liquid and eventually short-circuit the cell. Graphite anodes are safe but store far less energy per kilogram than lithium metal.

    A solid electrolyte solves both problems simultaneously. It’s not flammable, eliminating the fire risk. And it physically suppresses dendrite growth, making lithium metal anodes viable. Lithium metal anodes store roughly ten times the energy per gram that graphite does. Combined with high-voltage cathodes, this is what researchers call the “golden combination”—the pairing that lifts energy density from the 200–300 Wh/kg of current lithium-ion into the 400–500 Wh/kg range that changes the economics of electric vehicles, grid storage, and consumer electronics.

    Why it’s taking so long

    The interface problem. In a liquid electrolyte, the liquid conforms perfectly to the surface of the electrodes—every microscopic irregularity is contacted, every gap is filled. A solid electrolyte doesn’t do this. The solid-solid interface between the electrolyte and the electrode creates resistance from poor physical contact, chemical incompatibility that forms resistive layers, and mechanical stress from the volume changes that occur during every charge-discharge cycle. Lithium metal anodes expand and contract as lithium is deposited and stripped. After hundreds of cycles, the repeated expansion and contraction breaks the contact between the solid electrolyte and the anode, degrading performance and eventually killing the cell. Solving this requires advanced coating techniques, interface engineering, and novel electrode architectures—all of which are active areas of research, none of which are fully solved at manufacturing scale.

    The manufacturing problem. Conventional lithium-ion battery manufacturing is a mature, optimized, multi-trillion-dollar global industry. Solid-state batteries require entirely different manufacturing processes—different deposition techniques, different temperature profiles, different quality control parameters, different contamination tolerances. The sulfide electrolytes that Toyota and Samsung are pursuing are highly sensitive to moisture and require manufacturing environments with near-zero humidity. Building factories that can produce solid-state cells at the volumes and costs required for automotive deployment is a capital investment measured in billions of dollars, and the learning curve for scaling from pilot lines to gigafactory production is steep. Manufacturing costs currently sit at $400 to $800 per kilowatt-hour, compared to roughly $115/kWh for conventional lithium-ion in 2024. That’s a 4x to 7x premium that makes solid-state batteries economically viable only for niche, premium applications until manufacturing scale brings the cost curve down.

    The materials problem. There are three main electrolyte chemistries competing: sulfide, oxide, and polymer. Sulfides offer the highest ionic conductivity (meaning ions move through them fastest) but are unstable in air and moisture. Oxides are more stable but harder to manufacture as thin films and have lower conductivity. Polymers are easiest to process but generally require elevated temperatures to achieve adequate conductivity. Each chemistry has trade-offs, and the industry hasn’t converged on a single winner. Toyota is pursuing sulfides. Solid Power (BMW’s partner) started with sulfides. Samsung SDI is pursuing sulfides. The bet on sulfides is substantial, but the manufacturing sensitivity of the material is the primary bottleneck.

    Where everyone actually stands in 2026

    The field breaks into three tiers.

    Semi-solid-state batteries—hybrid cells with 5 to 15 percent liquid electrolyte retained—are already in vehicles. Chinese automakers Nio and IM Motors have shipped cars with semi-solid cells delivering 300 to 360 Wh/kg. These aren’t the full revolution, but they’re the bridge, and they’re real products in real cars being driven by real people. China’s official battery roadmap targets 350 Wh/kg liquid cells by 2025, 400 Wh/kg hybrid by 2030, and 500 Wh/kg true all-solid-state by 2035. China is also set to release its first national solid-state battery standard in July 2026.

    Pilot production of all-solid-state cells is underway or imminent at multiple companies. Toyota received production approval in October 2025. Samsung SDI promises 80 percent charge in nine minutes and 500 Wh/kg energy density, with mass production targeted for 2027. QuantumScape reports 80 percent capacity retention after 400 cycles at high charge rates in lab tests. Nissan is constructing a pilot factory in Yokohama. Dongfeng plans 350 Wh/kg mass production by late 2026. Statevolt’s 40 GWh gigafactory in the U.S. is projected to be operational in 2026, starting with semi-solid before transitioning to all-solid-state.

    Mass production at scale—the volumes needed to actually affect the automotive market—is not expected before 2028 at the earliest, with industry consensus placing large-scale commercialization closer to 2030. The global penetration rate of solid-state batteries is projected at roughly 0.1 percent in 2025, rising to about 4 percent by 2030 and approaching 10 percent by 2035. This is not a sudden disruption. It’s a decade-long ramp.

    What changes when they arrive

    The first-order effects are straightforward. Electric vehicles with 600 to 1,000 kilometers of range on a single charge, eliminating range anxiety as a barrier to adoption. Ten-minute fast charging, making EVs as convenient as gasoline cars at refueling stations. No fire risk, removing the safety concern that—while statistically rare in current EVs—generates disproportionate media coverage and consumer anxiety. Battery lifespans of 15 to 40 years, potentially outlasting the vehicle itself and enabling second-life applications in grid storage.

    The second-order effects are more interesting. A battery that lasts 40 years changes the economics of vehicle ownership fundamentally—you might keep the battery and replace the car around it. Grid-scale energy storage becomes dramatically more viable when the storage medium doesn’t degrade meaningfully over decades, which changes the economics of intermittent renewable energy. Aviation electrification becomes plausible for short-haul flights when energy density crosses the 400 Wh/kg threshold. Medical devices, military applications, space hardware, and extreme-environment operations all benefit from cells that operate reliably across wide temperature ranges without thermal management systems.

    The third-order effects involve supply chains. Solid-state batteries use less cobalt (or none, depending on cathode chemistry), which reduces dependence on the conflict-mineral supply chains in the DRC. They use more lithium metal, which shifts supply chain pressure toward lithium mining. They require new electrolyte materials—lithium sulfide, in Toyota’s case—which creates entirely new supply chains and potentially new geopolitical chokepoints. The race to secure solid-state battery materials is already underway, and the countries and companies that control the electrolyte supply chain will have leverage comparable to what China currently holds in rare earth processing.

    The honest timeline

    The pattern with solid-state batteries has been consistent for twenty years: the technology is always five years away. Toyota’s shifting deadlines—2020, 2023, 2026, now 2027–2028 for commercial vehicles—are representative of the entire field. The reasons for the delays are real and technical, not just corporate caution. The interface problem, the manufacturing problem, and the cost problem are genuine engineering challenges that don’t yield to deadline pressure.

    But the trajectory has changed. Semi-solid cells are in production vehicles today. All-solid-state pilot lines are running. Toyota has production approval. Samsung has published specs. The cost curve, while still far from competitive, is declining. The question is no longer whether solid-state batteries will work. It’s when they’ll be cheap enough, reliable enough, and produced at volumes high enough to displace the incumbent technology that powers essentially every EV, phone, and laptop on earth. The honest answer is: probably the early 2030s for mainstream automotive, with premium and niche applications arriving sooner.

    We cover solid-state batteries alongside fusion reactors, space elevators, and 21 other civilization-scale technology challenges across our Moonshot 2169 course—including why the most transformative battery technology of the 21st century has been “five years away” for two decades running.

  • Boston Dynamics vs. Tesla vs. Figure: The Humanoid Robot Race in 2026

    At CES 2026, Boston Dynamics unveiled the production version of Atlas—fully electric, 56 degrees of freedom, 50-kilogram lift capacity, autonomous battery swap—and won CNET’s “Best Robot” award. Every 2026 unit is already committed: they’re shipping to Hyundai’s Robotics Metaplant Application Center and Google DeepMind, with additional commercial customers planned from 2027. Korean securities firms valued Boston Dynamics between $21 and $28 billion, with bullish IPO projections reaching $88 to $103 billion. The company announced a strategic AI partnership with Google DeepMind and Toyota Research Institute. Its outgoing CEO, Robert Playter, said the goal is for Atlas robots to be “contextually aware of their environment and able to use their hands to manipulate any object.”

    That same month, Elon Musk announced that Optimus would go to the Moon. The previous year, he’d said it would go to Mars. Before that, he’d said Tesla would produce 5,000 to 10,000 Optimus units in 2025. The actual production number was reportedly in the hundreds. As of Q1 2026, Tesla confirmed that Optimus was still in an “R&D and learning phase” with no robots performing productive tasks in Tesla factories. The Optimus program lead since 2022, Milan Kovac, resigned in June 2025.

    Meanwhile, Figure AI closed a Series C round valuing the company at $39 billion—for a startup with only a few hundred commercial units deployed. Global robotics investment surpassed $10 billion in 2025. And the gap between valuations, promises, and actual robots doing actual work in actual facilities has never been wider.

    Three philosophies, one question

    The humanoid robot race in 2026 is not a single competition. It’s three companies making fundamentally different bets about what matters most.

    Boston Dynamics is betting on capability first, scale second. Atlas is the culmination of 13 years of continuous development, originally funded by DARPA for search-and-rescue operations. The old hydraulic Atlas could do backflips and run parkour courses. The new electric Atlas retains that dynamic agility while adding the manufacturability and reliability required for commercial deployment. Hyundai, Boston Dynamics’ majority shareholder, has committed $26 billion to U.S. manufacturing that includes a robotics factory capable of producing 30,000 units per year. The estimated price per unit is $140,000 to $150,000—enterprise-grade pricing for enterprise-grade performance. The target customer is a Fortune 500 manufacturer, not a consumer.

    Tesla is betting on scale first, capability second. Optimus is designed from the outset for mass production, leveraging Tesla’s automotive supply chain, manufacturing expertise, and AI infrastructure (the same Full Self-Driving platform that powers its vehicles). Musk’s target price is $20,000 to $30,000—deliberately “less than a car.” At that price point, if the robot can perform useful tasks, the addressable market is essentially every warehouse, factory, and eventually every household on earth. The problem is the “if.” Every public Optimus demonstration has been criticized for signs of remote human control. Tesla has never held a fully autonomous public demonstration without controversy. The V2.5 iteration, revealed in late 2025, improved the cosmetic design but reviewers described it as underwhelming in function—slow voice command response, tentative motion, awkward pauses. Cosmetic refinement outpacing demonstrable capability is not the trajectory you want if your thesis depends on the robot actually working.

    Figure AI is betting on speed and capital. Founded in 2022, the company raised over $1.6 billion and reached a $39 billion valuation in roughly three years—a pace that would be remarkable even by Silicon Valley standards. Figure’s approach combines a hardware platform (the Figure 02, estimated at over $100,000 per unit) with aggressive AI integration, including a partnership with OpenAI for natural language interaction and task understanding. The company has deployed a few hundred units commercially and is positioning itself as the startup most likely to bridge the gap between research platform and scalable product. Whether a three-year-old company can compete with Boston Dynamics’ 30 years of locomotion research and Tesla’s manufacturing infrastructure is the open question, but the capital markets are clearly betting that the AI component—making the robot understand what you want it to do—matters more than the hardware component.

    What the robots can actually do right now

    This is where the marketing and the engineering diverge most sharply.

    Atlas can walk, run, jump, recover from dynamic perturbations, manipulate objects with precision, and navigate unstructured environments. The CES 2026 demonstration showed car part sequencing and factory component handling. It was remotely operated during the demo, but Boston Dynamics stated the commercial version will be fully autonomous. The company has a deployment track record with Spot (the quadruped) and Stretch (the warehouse robot) that provides credibility for its commercialization claims. Atlas’s 56 degrees of freedom, water resistance, and extreme temperature tolerance make it the most physically capable humanoid robot in production.

    Optimus can walk with an improved heel-to-toe gait, perform basic pick-and-place operations, and handle simple household tasks in staged demonstrations—stirring a pot, sweeping, vacuuming. These are legitimate capabilities, but they’re a long distance from productive factory work. Tesla has not published operational metrics like cycle times, task completion rates, or hours of autonomous operation. Independent reporting suggests the robots deployed internally at Tesla factories are in a learning phase rather than performing useful labor. Musk acknowledged in Tesla’s Q4 2025 earnings that the robots are “not doing useful work” yet. Consumer sales are targeted for late 2027.

    Figure 02 has demonstrated warehouse picking and packing tasks, object manipulation, and natural language interaction through its OpenAI integration. The company has deployed units at BMW’s Spartanburg manufacturing plant and other commercial sites. The demonstrations are impressive but limited in scope, and the deployed fleet numbers in the hundreds—enough for pilot programs, not enough for operational conclusions about reliability or economics.

    The China factor

    The comparison that the American companies probably don’t want you making is with Chinese humanoid robot manufacturers, who are approaching the problem from a different angle entirely. Unitree’s humanoid models emphasize agile maneuvers at accessible price points. UBTECH’s Walker series has demonstrated autonomous battery swapping for continuous 24/7 operation—a practical advantage in factory settings where uptime matters more than acrobatics. BYD, the EV manufacturer, targeted 1,500 humanoid robot units in 2025 and is scaling to 20,000 by 2026.

    Chinese firms are pursuing narrow, production-centric optimization: rapid iteration, manufacturability, duty-cycle engineering, and lower unit costs. Their emphasis is less on demonstrating a spectacular generalist and more on producing reliable, maintainable machines for specific operational roles. That approach—boring, incremental, manufacturing-focused—is exactly what scaled industrial deployment actually requires, and it’s the approach most likely to produce the first humanoid robot that earns its keep on a factory floor without a press release attached to every shift.

    Manufacturing costs across the industry dropped roughly 40 percent from 2023 to 2024, falling from $50,000–$250,000 per unit to $30,000–$150,000, according to Goldman Sachs. That trajectory, if it continues, brings the economics of humanoid robots into range for mainstream industrial deployment by the late 2020s regardless of which specific company gets there first.

    The honest scorecard

    On technical capability in 2026: Atlas wins. It’s not close. Thirty years of locomotion research, DARPA funding, the transition from hydraulic to electric, Google DeepMind AI integration, and a parent company willing to build a 30,000-unit factory give it advantages that no competitor can replicate in the short term.

    On manufacturing potential: Tesla wins, theoretically. No company on earth has more experience producing complex electromechanical systems at scale. If Optimus reaches the point where it can perform useful work autonomously, Tesla’s production infrastructure is unmatched. The problem is that “if,” and the gap between Musk’s production targets and actual output has been growing, not shrinking.

    On capital and velocity: Figure AI wins. A $39 billion valuation and aggressive AI partnerships give it the resources and talent to move fast. Whether moving fast in robotics translates to a durable product—as opposed to a series of impressive demos—is unproven.

    On realistic near-term deployment: Boston Dynamics wins again. It’s the only company in this comparison with a production robot shipping to commercial customers in 2026, backed by a parent company with a $26 billion manufacturing commitment and a track record of commercializing previous robots (Spot, Stretch) that actually operate in the field.

    The humanoid robot that reshapes industry probably isn’t the one that does the best backflip or gets the highest valuation. It’s the one that shows up to work on a random Tuesday, completes its task list without human intervention, and does it again the next day for less than the cost of hiring a person. None of the three companies in this comparison has demonstrated that yet. The race isn’t over. It arguably hasn’t started—because the real competition begins when the robots stop performing and start producing.

    We cover the engineering, AI integration, and commercial deployment of humanoid robots across 24 lectures in our Humanoid Robots & Drones course—including why the company that wins the humanoid robot race might not be the one making the best robot.

  • Can BCIs Treat Depression? The Science of Neural Stimulation for Mental Health in 2026

    In December 2025, the FDA approved the first at-home brain stimulation device for depression. The Flow FL-100, made by a Swedish company called Flow Neuroscience, is a headset that delivers low-level electrical current to the prefrontal cortex—the part of the brain involved in mood regulation and stress response—for 30 minutes at a time. The clinical trial that earned the approval showed 58 percent of patients reaching remission after 10 weeks. The device will be available by prescription in the United States by mid-2026, at a retail price between $500 and $800. Over 55,000 patients have already used it across Europe, the UK, Switzerland, and Hong Kong.

    That’s the accessible end of the spectrum. At the other end—surgically implanted electrodes delivering personalized, closed-loop electrical stimulation directly to deep brain structures—the science is more dramatic, more preliminary, and considerably more difficult to scale. Both approaches share a foundational premise that would have sounded like science fiction twenty years ago: that depression, at least in some patients, can be treated by altering electrical activity in specific brain circuits rather than flooding the entire brain with neurotransmitter-modifying drugs. The question in 2026 is not whether neural stimulation works for depression. It’s how precisely it needs to work, for whom, and at what cost—financially, surgically, and ethically.

    The spectrum of stimulation

    Neural stimulation for mental health spans a range of invasiveness, precision, and evidence quality. Understanding where each technology sits on that spectrum matters more than any individual headline.

    Transcranial direct current stimulation—tDCS—is the least invasive. A device sends a weak electrical current (typically 1 to 2 milliamps) through electrodes placed on the scalp. The current modulates the excitability of neurons in the targeted region without directly triggering them to fire. The Flow device uses this approach. The evidence base is mixed: some trials show clear benefits over placebo, others find little difference. The FDA approval was based on a 174-participant trial published in Nature Medicine. The effect is real but modest—this is not a cure, it’s a tool, and it works better in some patients than others for reasons that aren’t fully understood.

    Transcranial magnetic stimulation—TMS—uses magnetic pulses to induce electrical currents in specific brain regions. It’s been FDA-approved for treatment-resistant depression since 2008 and is administered in clinics, typically over multiple sessions spanning weeks. Repetitive TMS targeting the left dorsolateral prefrontal cortex has the strongest evidence base among non-invasive brain stimulation approaches. An accelerated protocol called Stanford Neuromodulation Therapy, developed at Stanford and published in 2022, compressed the treatment course into five days of intensive stimulation sessions and achieved remission rates approaching 80 percent in a small trial of treatment-resistant patients. The protocol uses brain imaging to personalize the stimulation target for each patient—a significant departure from one-size-fits-all approaches.

    Vagus nerve stimulation—VNS—involves surgically implanting a device that electrically stimulates the vagus nerve in the neck, which sends signals to brain regions involved in mood regulation. It’s been FDA-approved as an adjunctive treatment for treatment-resistant depression since 2005. Response rates are modest and build slowly over months to years. Non-invasive vagus nerve stimulation devices, which stimulate the nerve through the skin of the ear or neck, are being investigated but lack the same evidence base.

    Deep brain stimulation—DBS—is the most invasive: surgeons implant electrodes directly into specific brain structures and deliver electrical impulses through a battery-powered device implanted in the chest. DBS is FDA-approved and well-established for Parkinson’s disease, with over 12,000 patients receiving the treatment annually. For depression, it remains experimental—and the history of DBS for depression is one of the most instructive stories in psychiatric neuroscience about the distance between a promising concept and a working treatment.

    The DBS depression story

    The modern era of DBS for depression began in the early 2000s, when neurologist Helen Mayberg identified a brain region called the subcallosal cingulate—also known as Brodmann area 25—as a key node in the neural circuits underlying depression. In a landmark 2005 study, Mayberg and colleagues implanted DBS electrodes targeting this region in six patients with severe, treatment-resistant depression. Four of six experienced sustained remission. The results were dramatic enough to generate enormous excitement and multiple larger clinical trials.

    Those trials, conducted through the late 2000s and 2010s, produced highly variable results. A major randomized controlled trial sponsored by St. Jude Medical (now Abbott) was halted in 2013 after a futility analysis suggested the treatment was unlikely to show significant benefit over sham stimulation. The failure was attributed to multiple factors: imprecise electrode targeting, continuous rather than responsive stimulation, heterogeneity in the depression circuits of different patients, and the fundamental problem that depression doesn’t appear to have a single anatomical locus that’s the same in everyone. What worked in Mayberg’s initial patients didn’t generalize to the broader population with the same stimulation parameters.

    The insight that emerged from these failures was that DBS for depression probably can’t be standardized the way DBS for Parkinson’s is. Depression circuits vary between individuals. The biomarker that indicates when stimulation is needed varies between individuals. The brain target where stimulation is most effective varies between individuals. A treatment that works has to be personalized at every level.

    The UCSF closed-loop breakthrough

    This is where the UCSF trial becomes significant. In October 2021, Katherine Scangos, Edward Chang, and Andrew Krystal published a case report in Nature Medicine describing a fundamentally different approach to DBS for depression. Their patient, a 36-year-old woman known as Sarah, had childhood-onset severe depression that had been unresponsive to multiple antidepressant combinations and electroconvulsive therapy. Her depression rating score was 36 out of 54 on the standard scale.

    The team first implanted ten temporary electrodes across Sarah’s brain for a 10-day mapping period. They stimulated each brain region individually while Sarah rated her symptoms, identifying which targets relieved which specific depression symptoms. They simultaneously recorded continuous neural activity while Sarah completed symptom ratings, identifying a personalized biomarker: elevated gamma-band activity in her amygdala correlated with her most severe depressive states.

    They then implanted a NeuroPace RNS System—a device originally developed and FDA-approved for epilepsy—with one electrode lead in the amygdala to sense the biomarker and another in the ventral capsule/ventral striatum to deliver stimulation when the biomarker was detected. The system delivered a tiny pulse—one milliamp for six seconds—only when it detected the neural signature of an oncoming depressive state. Closed-loop. Responsive. Personalized.

    The result was rapid and sustained improvement. Sarah described the initial stimulation as “the most intensely joyous sensation.” Over subsequent months, the device continued to manage her depression in real time. She reported that intrusive depressive thoughts still arose but “it’s just… poof… the cycle stops.” Fifteen months after implantation, the improvement had held.

    The UCSF team has since enrolled additional patients in the trial and expanded to bipolar depression. Mount Sinai performed the first DBS implant for depression as part of a separate clinical trial in March 2025. STAT News identified brain implants for mental health as one of the top three BCI trends to watch in 2026. An IEEE Spectrum analysis published in August 2025 described AI-enhanced DBS that could predict depressive relapses before they occur and adjust stimulation parameters proactively.

    What this doesn’t mean yet

    The honest assessment requires a few buckets of cold water. Sarah is a single patient. An n-of-1 case report, however dramatic, does not constitute evidence that closed-loop DBS will work for depression broadly. The UCSF team has said as much explicitly: “We need to look at how these circuits vary across patients and repeat this work multiple times.” The treatment requires brain surgery—two separate procedures in Sarah’s case. The NeuroPace device is FDA-approved for epilepsy, not depression; its use in the UCSF trial was under an investigational device exemption. FDA approval for DBS as a depression treatment is, in the researchers’ own estimation, still far down the road.

    The earlier DBS trials failed not because the concept was wrong but because the implementation wasn’t personalized enough. Whether the closed-loop, biomarker-driven approach solves that problem at scale—across the enormous heterogeneity of depression as a diagnosis—is an empirical question that will take years and many more patients to answer.

    More than 20 million American adults live with depression, a 60 percent increase over the past decade. Approximately one-third don’t respond adequately to antidepressant medications. For most of those patients, the relevant intervention in 2026 is not an implanted electrode—it’s better access to existing treatments, including TMS and potentially the new at-home tDCS devices. The $500 Flow headset and the surgically implanted closed-loop DBS system represent opposite ends of a continuum, and the clinical reality for most patients with treatment-resistant depression sits somewhere in the middle, where the options are expanding but the solutions are still imperfect.

    The trajectory, though, is unmistakable. The field is moving from treating depression as a chemical imbalance—the serotonin model that dominated psychiatry for decades and has been increasingly questioned—toward treating it as a circuit disorder, where specific patterns of electrical activity in identifiable brain networks produce specific symptom clusters, and those patterns can be detected, modulated, and corrected. That reframing, more than any individual device, is the development worth watching.

    We cover neural stimulation for depression alongside the full landscape of brain-computer interfaces—from motor prosthetics to speech restoration to sensory augmentation—across 48 lectures in our Neuroprosthetics course. If the shift from treating depression as chemistry to treating it as circuitry changes how you think about mental health, the course goes deep on the neuroscience and engineering behind every approach on the spectrum.