The tech industry was caught off guard by the rapid rise of AI following the launch of ChatGPT. Power grids, cooling systems, and data centers are now being overhauled because they were not built to handle the unprecedented demand for computing power. According to United Nations researchers, AI could double the energy and water consumption of data centers by 2030. If the industry had anticipated this surge, it might have designed infrastructure to handle it. Quantum computing, still in its early stages, could be the solution if it is developed with these challenges in mind. For companies that operate large-scale AI systems, the key metric is how much useful computation they can get for each dollar invested. This total cost of ownership is largely determined by the cost of building the infrastructure and the amount of energy it uses. While these companies have access to capital, they are currently facing a severe shortage of power. This has led to the idea of building data centers in space, where power might be more readily available. Even if the AI industry slows down, the cost of energy and infrastructure will remain critical factors. In the quantum computing field, the focus should shift from the number of qubits—basic units of quantum information—to how much computing power each watt of energy provides. However, quantum computers are still marketed based on qubit counts, which don't fully reflect their economic value. What really matters at scale is the amount of computation per watt a system can deliver, which isn’t solely determined by the number of qubits. For example, superconducting qubit processors, one of the most promising quantum computing platforms, have struggled to move beyond around 100 qubits for nearly a decade. This is because most of the chip's surface is used for wiring that controls and reads the qubits, rather than for the qubits themselves. One common solution is to link multiple smaller processors together, but this approach leads to inefficiencies. In classical computing, this isn't a major issue, but in quantum computing, it creates exponential problems. The effort to manage the system grows faster than the computing power it provides, increasing the overall energy cost and reducing the efficiency of the system. The concept of an experience curve, where the cost of a product decreases as more of them are produced, has driven down the prices of solar panels and batteries. Quantum computing could follow a similar path, but only if it is manufactured at high volumes. Current price estimates suggest that a quantum computer with a million qubits would be extremely expensive—between 100 billion and a trillion dollars. To make quantum computing economically viable, the cost per qubit must decrease significantly, which is achievable but not yet realized through laboratory experiments alone. The evolution of the transistor is a prime example of how costs can be reduced through mass production and standardization. Today's computer chips contain billions of transistors, each costing just a fraction of a cent. This reduction in cost came from decades of mass production and continuous technological improvements. Quantum computing needs a similar approach, with specialized companies working together to build different parts of the system. This Quantum Open Architecture—where processors, cooling systems, and control electronics are developed separately—could drive the economies of scale needed for quantum computing to become practical. As quantum computing moves from a scientific experiment to an engineering discipline, there are many impressive one-time demonstrations by startups, but few companies are building the necessary supply chain. Quantum computers have the potential to revolutionize fields like drug discovery, materials science, and machine learning. They could also help reduce the strain on power grids by performing complex calculations more efficiently. However, achieving this potential will require focusing on the economic factors that drive real-world applications. The number of qubits alone doesn’t determine success. Building a robust supply chain will take years, and the availability of industrial fabrication, standardized parts, and processors with significantly more computing power per watt will depend on the choices made in the coming years.