Will AI and robots really make money useless? Last updated: 31 July 2026 In our robotics market deck , you will find everything you need to understand the market SUMMARY No, AI and robots are unlikely to make money useless. They could make basic living much cheaper and paid work less necessary, but money will survive wherever land, energy, human attention, ownership and access remain scarce. The idea is no longer just science fiction because two changes are happening together. Generative AI has made useful digital intelligence extremely cheap, while robots have started doing measurable paid work in factories and warehouses. That abundance is real but narrow. A cheap AI answer still depends on expensive chips, data centres, electricity, cooling, fibre and financing, so the visible service can approach zero cost while the underlying system remains highly capital-intensive. AI productivity also looks far stronger at task level than at company or economy level. Controlled studies show large gains in coding, support and marketing, while most executives still report little measurable effect on total employment or firm-wide productivity. Job pressure is arriving unevenly. Junior white-collar hiring appears more exposed than experienced work, which suggests that AI may damage entry routes before it eliminates whole occupations. Humanoid robots have crossed an important line from demonstrations to commercial deployments, but the scale gap is enormous. Hundreds of units and planned factories are not yet evidence that billions of reliable general-purpose robots can be produced economically. Automation will push prices down most sharply for digital and standardized products. Housing, food, healthcare and infrastructure will remain expensive because labor is only one part of their cost. Scarcity will probably move rather than disappear. Prime locations, grid capacity, trusted human attention, rare experiences and social status remain limited even when ordinary goods become plentiful. Ownership may be the decisive issue. Machine output initially belongs to the companies, investors and governments that own the models, data centres and robot fleets, so technical abundance does not automatically become shared abundance. The transition could be rougher than the destination. Wages and job security can weaken faster than tax systems, public services and income guarantees can adapt, making money more urgent for some households before automation makes life cheaper. The most plausible outcome is a two-layer economy: a high-quality baseline of intelligence, education, healthcare, transport and manufactured goods becomes cheap or publicly guaranteed, while money continues to allocate scarce property, premium services, rare experiences and control over productive assets. This market map, featured in our robotics market deck , highlights top companies and startups in the robotics market Why are people seriously talking about money becoming useless now? Yes, the idea deserves a serious look now because cheap AI, mass adoption and early commercial humanoids are arriving at the same time. Elon Musk has pushed the strongest version of the claim. At the World Economic Forum, he described AI and robotics as the route to “abundance for all,” predicted more robots than people and argued that machines could eventually satisfy almost every request humans can think of. More recently, he told The Economist that money could stop mattering within about a decade. That forecast would have sounded almost purely philosophical before generative AI. Today, Stanford’s 2026 AI Index estimates that generative AI reached 53% adoption within three years, faster than the early spread of the internet or personal computers. It also estimates that the value U.S. consumers receive beyond what they pay rose from 112billionto112 billion to 172 billion in one year, while many leading tools remained free or close to free. Robotics is moving too, although much more slowly. Figure says its humanoid loaded more than 90,000 parts during 1,250 hours at BMW and contributed to 30,000 vehicles. Agility Robotics says Digit has moved more than 100,000 totes in a commercial warehouse deployment. These are narrow jobs, but they are paid work inside real operations rather than stage demonstrations. So the timing of the question makes sense. Digital intelligence is becoming abundant quickly, and useful physical robots have finally appeared. Treating those early achievements as proof that every form of scarcity is about to vanish would still be a huge leap. What would it actually mean for money to become useless? Money would become useless only if people no longer needed it to obtain scarce things, compare choices or save a claim on future goods. In normal life, money solves three simple problems. It lets us trade without swapping one specific object for another. It gives us a common way to compare the price of a meal, a house and an hour of work. It also lets us save purchasing power for later. AI and robots could weaken the first problem by making many products extremely cheap. They could weaken the second if basic services became universally available. Yet the third remains whenever people want to save for something scarce tomorrow, and all three remain wherever demand exceeds supply. That leaves four very different futures that are often described with the same “post-money” label. Only the last one would truly make money useless. Possible future What changes Does money disappear? AI becomes nearly free Everyone gets cheap digital help No Work becomes optional People receive income without needing a job No Basic living is guaranteed Housing, healthcare, food or transport are provided Money matters less for survival Scarcity largely disappears Almost every desired good is available to everyone Money may lose most of its role If you want more recent data on this point, please see our latest robotics market report . As this chart shows, and as featured in our robotics market deck , search interest in robot costs has increased significantly Is cheap AI already creating real abundance? Cheap AI is already creating a narrow but meaningful kind of abundance in information, software and creative work. Stanford’s latest AI Index captures the scale of the change. It estimates that 88% of surveyed organizations now use AI somewhere, while 70% use generative AI in at least one business function. Consumer value rose 54% in a year, and the median value reported by users tripled. Millions of people can now get writing help, translations, basic coding, tutoring and image generation at little or no direct cost. This counts as real abundance because digital output can be copied repeatedly. Once a model and its infrastructure exist, serving one more short answer is far easier than building one more house or manufacturing one more car. The expensive part has moved behind the screen. The International Energy Agency says five large technology companies spent more than 400billiononcapitalexpenditurein2025,withanotherlargejumpexpected.TheirAIdatacentrecapacitymorethantripledin18months.Usersseeacheapanswer;underneathitsitchips,coolingsystems,powerplants,fibrenetworksandenormousfinancingneeds.Digitalabundanceisalreadyreal,butitcurrentlystopsatinformationandexpertise.Housing,food,energyandmanufacturedgoodsstillliveinamuchmoreexpensiveworld.IsAImakingthewholeeconomymoreproductiveyet?AIisdeliveringstronggainsinselectedtasks,buttheeconomywideproductivityboomisstillmostlyaheadofus.Thebestexperimentskeepfindingasimilarpattern.AIworksespeciallywellwhenthetaskisstructuredandtheresultiseasytocheck.AnNBERstudyof5,179supportagentsfounda14400 billion on capital expenditure in 2025, with another large jump expected. Their AI data-centre capacity more than tripled in 18 months. Users see a cheap answer; underneath it sit chips, cooling systems, power plants, fibre networks and enormous financing needs. Digital abundance is already real, but it currently stops at information and expertise. Housing, food, energy and manufactured goods still live in a much more expensive world. Is AI making the whole economy more productive yet? AI is delivering strong gains in selected tasks, but the economy-wide productivity boom is still mostly ahead of us. The best experiments keep finding a similar pattern. AI works especially well when the task is structured and the result is easy to check. An NBER study of 5,179 support agents found a 14% productivity increase, with much larger gains for inexperienced workers. Stanford’s latest review reports gains around 26% in software development and up to 50% in some marketing-output experiments. Company-wide results look far smaller. A 2026 NBER survey covering executives across several countries found that more than 90% reported no employment effect from AI during the previous three years, while 89% reported no productivity effect. Across all firms, including the minority seeing gains, the estimated productivity lift was about 0.29%. When economists zoom out from individual tasks, the numbers shrink again. The OECD estimates that AI could eventually add around 0.25 to 0.6 percentage points to annual productivity growth in advanced economies. Daron Acemoglu’s more cautious model puts the ten-year productivity gain below 0.55% in total. Those forecasts still represent meaningful growth. They remain far short of unlimited production. Right now, AI is spreading across many industries, with its impact slowed by training, bad data, unreliable outputs and the difficulty of redesigning whole organizations around it. This chart, featured in our robotics market deck , shows annual venture capital investment in robotics startups Is AI actually replacing workers today? AI is starting to squeeze particular hiring pipelines, although mass replacement has not appeared in the overall employment data. The International Labour Organization finds that one in four workers holds a job with some exposure to generative AI. Most of those jobs contain tasks that can be changed or accelerated, while relatively few can be fully handed to a model from beginning to end. That difference between exposed tasks and replaceable workers is easy to miss. The freshest evidence shows pressure at the edges. Stanford reports that employment among software developers aged 22 to 25 fell nearly 20% from 2024, a much sharper change than for older developers. One-third of organizations in its survey expect AI to reduce their workforce over the coming year, especially in software engineering, service operations and supply chains. At the same time, the NBER’s global firm survey found that nine in ten executives had seen no employment effect so far. Anthropic’s workplace research also finds that AI is used heavily in parts of coding and writing, but rarely completes every task inside an occupation. Pressure is landing first on junior white-collar work. A broad job apocalypse remains a forecast. Today’s evidence shows some tasks and entry-level roles being squeezed long before whole occupations disappear. If you want more recent data on this point, please see our latest robotics market report . Can robots really take over the physical economy? Robots can automate a large share of physical work, but today’s successful machines still depend on predictable spaces and tightly defined jobs. Industrial robotics is already a mature industry. The International Federation of Robotics recorded 542,000 factory robot installations in 2024, marking the fourth straight year above 500,000. China installed 54% of the global total, more than every other country combined. Professional service-robot sales reached almost 200,000 units in the same year. These machines weld, package, inspect, carry and place objects with impressive speed. Their productivity often comes from arranging the factory around them. Parts arrive in known positions, routes stay clear and safety rules limit unexpected contact. Humanoids are beginning to work outside traditional cages. Figure’s BMW deployment handled sheet-metal parts through ten-hour shifts. Agility’s Digit moves totes between warehouse equipment. Both jobs use the human shape to navigate buildings designed for people, but neither robot currently behaves like a general employee who can switch freely between cleaning, repairing machinery, caring for someone and cooking lunch. Automation is moving through the physical economy one layer at a time. Repetitive factory and warehouse work comes first, while construction sites, farms, hospitals and homes bring far more variation, weather, fragile objects and human responsibility. Area Current reality Main obstacle to wider automation Factory work Hundreds of thousands of new robots each year Reprogramming and flexible handling Warehouses Large mobile-robot fleets and early humanoid deployments Handling many object shapes safely Construction and farming Useful specialized machines Unstructured outdoor conditions Hospitals and care Robots assist with surgery, logistics and monitoring Safety, trust and human judgment Homes Cleaning and mowing are common General manipulation in messy spaces This chart, featured in our robotics market deck , breaks down Figure’s playbook in robotics Are humanoid robots anywhere close to the scale this vision requires? Humanoid robots are scaling faster than before, but they remain several orders of magnitude below a world with one robot for every person. Figure says it produced more than 350 third-generation humanoids and increased its factory rate from one robot per day to one per hour in less than four months. Agility’s Oregon factory is designed for peak annual capacity of 10,000 Digits. Tesla says its first Optimus line is being designed for one million robots per year, followed by a longer-term Texas line designed for ten million. The words “designed for” matter. Figure has hundreds of new robots, Agility has commercial units in specific facilities, and Tesla’s million-unit line remains a production plan. None has yet shown consumer-scale output, multi-year reliability or economics comparable with cars and appliances. The gap is enormous even when we compare the vision with the entire industrial robot industry. Adding eight billion robots over 20 years would require average production of 400 million a year. That is roughly 740 times the latest annual installation rate for all factory robots, not merely humanoids. Production can rise exponentially, so today’s rate does not set a permanent ceiling. Car manufacturing proves that complex machines can eventually reach tens of millions of units a year. Still, humanoids performing real tasks today tells us very little about whether billions of general robots will arrive soon. Would AI and robots make most products almost free? AI and robots will push many prices down, but physical products will keep substantial costs after labor becomes cheaper. Automation attacks several expensive parts of production. AI can speed up design, purchasing, scheduling and quality control. Robots can run longer shifts, reduce defects and perform repetitive work. Autonomous transport can lower some delivery costs. Materials, land, machines, energy, financing and maintenance remain on the bill. A robotic construction crew still needs a plot of land, concrete, steel, permits and a grid connection. An automated farm still needs water, fertilizer, equipment and usable soil. A drug discovered by AI still needs trials, factories, regulation and distribution. The AI industry itself shows why cheap output can coexist with massive costs. The latest International Energy Agency figures put data-centre electricity use at 485 terawatt-hours in 2025 and project roughly 950 terawatt-hours by 2030. AI-focused centres grew much faster than the overall category. Meanwhile, technology companies are committing hundreds of billions of dollars to the infrastructure that makes low-cost AI services possible. Prices can fall dramatically without reaching zero. Expect the strongest declines where the final product is digital or standardized, and the weakest where location, materials, regulation or personal care dominate. Product or service Likely price effect What keeps it from becoming free Basic translation, coding and digital content Very large decline Compute and premium quality Standard manufactured goods Large decline Materials, factories and transport Food Moderate to large decline Land, water, fertilizer and weather Housing construction Moderate decline Land, permits, materials and utilities Healthcare Uneven decline Regulation, liability and human care Prime property and unique experiences Small decline Supply is deliberately or naturally limited If you want more recent data on this point, please see our latest robotics market report . This chart, featured in our robotics market deck , shows annual funding in robotics startups What would still be scarce in a robot economy? Land, energy, infrastructure, human attention and social position would remain scarce even in a highly automated economy. Land provides the easiest example. Robots can build more homes, but they cannot produce unlimited apartments overlooking Central Park or beside a protected beach. Better transport may create new desirable areas, yet some locations will always be preferred over others. Energy and infrastructure are already becoming visible limits. The International Energy Agency says global data-centre electricity consumption grew 17% in 2025, while AI-focused centres grew 50%. It also reports shortages in high-bandwidth memory and long delays for grid connections, transformers, turbines and other equipment. Efficiency is improving quickly, but usage is growing even faster. Human attention has a harder limit. One famous surgeon, trusted adviser, performer or close friend still has 24 hours in a day. AI can copy knowledge and style, but it cannot make one particular person personally available to everyone. Status creates scarcity even when the underlying product is easy to make. If everyone owns an excellent robot-made watch, distinction moves toward the rare watch, the original artwork or the invitation that only a few people receive. Automation can raise the standard floor while competition continues above it. Could energy or computing power replace money? Energy and compute may become valuable rationing units, but they are too uneven and specialized to replace money across the economy. The idea sounds appealing because every AI model and robot ultimately needs energy. Musk himself currently describes electricity as the main limit on AI deployment. The International Energy Agency’s latest work supports the concern: data-centre demand is rising quickly, advanced server racks are becoming much more power-dense and new grid capacity takes years to build. A kilowatt-hour still has a different value depending on where and when it is available. Electricity beside a data centre during a shortage is more useful than the same amount produced far away when transmission lines are full. Storing and moving it also costs money. Compute varies even more. A simple text query and an advanced video or agentic task can differ in energy use by hundreds or thousands of times. Chips also differ in speed, memory and software support. One generic “compute credit” would hide all those differences. Governments or companies could issue energy allowances, compute credits or robot hours. Once people can save, exchange and price those credits, they start behaving like specialized currencies. The label changes, while the basic role of money survives. This chart, featured in our robotics market deck , compares the main business model options for warehouse AMR robotics providers If robots create the wealth, who actually gets it? The owners of AI systems, data centres and robot fleets receive the first claim on their output, so abundance will not spread automatically. Current ownership is heavily concentrated. In the Federal Reserve’s latest Distributional Financial Accounts, the wealthiest 10% of U.S. households hold about 48.2 trillion of the country’s 55.1trillionincorporateequitiesandmutualfundshares.Wecalculatetheirshareatroughly8755.1 trillion in corporate equities and mutual-fund shares. We calculate their share at roughly 87%. The bottom half holds about 590 billion, close to 1%. That starting point shapes the automation story. When a company replaces paid labor with machines, wages may fall while profits and the value of the machines rise. Workers benefit broadly only when they also own those assets, receive higher wages in complementary jobs, pay lower prices, or gain access through taxes and public services. The IMF’s recent inequality res