Can engineering innovations make the "artificial sun" smaller and more controllable? Controlled nuclear fusion is widely regarded as the "crown jewel" of energy technology. With abundant fuel sources and near-zero carbon emissions during operation, fusion promises a virtually limitless supply of clean energy. Since the 1950s, scientists have pursued various methods to achieve the extreme conditions required for fusion. Today, the field is witnessing rapid progress, with global fusion industry growth accelerating dramatically. According to the Fusion Industry Association’s (FIA) 2025 Global Fusion Industry Report, released in July 2025, the fusion sector has seen explosive expansion over the past five years. The report reveals that there are now 53 fusion companies worldwide, with total investment reaching $9.77 billion—five times higher than in 2021. Notably, $2.64 billion was invested in just the past year. To put the potential of fusion into perspective: generating 1,000 megawatts of electricity would require about 1.4 million tons of coal in a conventional power plant. In contrast, under ideal conditions, a fusion reactor would need only approximately 0.6 tons of deuterium-tritium fuel—equivalent to about 1,200 standard 330ml soda bottles. This highlights the immense energy density of fusion. Among all fusion reactions, the deuterium-tritium (D-T) reaction remains the most accessible and well-studied path. Deuterium is easily extracted from seawater, while tritium can be bred through neutron interactions with lithium. To initiate D-T fusion, the fuel must be heated to around 100 million degrees Celsius—high enough for atomic nuclei to overcome electrostatic repulsion and fuse, releasing a helium nucleus and a neutron, along with massive energy. Traditional heating methods rely on external energy input—similar to "boiling water"—using techniques like neutral beam injection or ion cyclotron heating to inject high-energy particles into the plasma. Since the 1990s, major magnetic confinement devices such as JET (Joint European Torus) and TFTR (Tokamak Fusion Test Reactor) have achieved short bursts of tens of megawatts of fusion power, proving the scientific feasibility of fusion energy. More recently, the U.S. National Ignition Facility (NIF) has achieved net energy gain multiple times using laser inertial confinement, producing fusion energy equivalent to one kilowatt-hour. Meanwhile, top journals like Nature and Science have published breakthroughs demonstrating that artificial intelligence can significantly enhance plasma confinement and magnetic field control—marking a new frontier in fusion research. However, achieving "ignition" is not enough. Sustained, stable energy production remains a major engineering challenge. While scientific feasibility has been demonstrated, real-world commercial fusion requires solving complex issues related to plasma confinement, heating efficiency, and system durability. Fusion is now a key battleground in global technological competition. The U.S. views fusion as central to maintaining its leadership in the 21st century. In 2023, the U.S. Department of Energy awarded $46 million to eight fusion startups under its milestone-driven development program. China has also taken strong policy steps: the recently passed China Atomic Energy Law (September 2024) explicitly encourages research, development, and innovation in controlled nuclear fusion. Beyond government support, private companies are driving rapid innovation. Fusion technologies fall into two main categories: magnetic confinement (including tokamaks, stellarators, and field-reversed configurations) and laser inertial confinement. Commercialization depends on three key factors: confinement performance, magnetic field strength, and device size. Historically, improving confinement and magnetic field strength has been difficult, forcing engineers to rely on massive devices. The International Thermonuclear Experimental Reactor (ITER), a 30-meter-wide, 30-meter-tall tokamak built by seven nations, has a budget of $20 billion and has yet to be fully completed after starting construction in 2010. But recent advances suggest that engineering innovations could dramatically shrink fusion devices. Companies like Commonwealth Fusion Systems (CFS), Helion Energy, and Tokamak Energy are pioneering compact, high-performance designs. CFS, spun off from MIT in 2018, uses high-temperature superconducting magnets in its SPARC tokamak. In August 2025, it raised $863 million in a new funding round led by NVentures (NVIDIA’s venture arm), Google, and Breakthrough Energy Ventures (founded by Bill Gates). The company has now raised around $3 billion and signed a $1 billion fusion power agreement with Italy’s Eni. Helion Energy, founded in 2013, uses a field-reversed configuration that enables direct electricity generation. In 2021, its Trenta device achieved 100 million degrees Celsius, and the company claims its fusion generator can run continuously for 16 months. In 2023, it partnered with Microsoft to deliver fusion power by 2028. In July 2025, construction began on its fusion power plant in Washington State. It is currently developing its seventh-generation prototype, Polaris. Helion raised $425 million in a Series F round in January 2025, with support from Sam Altman (OpenAI CEO) and SoftBank Ventures, bringing its valuation to $5.425 billion. Tokamak Energy, based in the UK and founded in 2009, focuses on spherical tokamaks and high-temperature superconducting magnets. Its ST40 device achieved 100 million degrees Celsius in 2022. The company has raised $335 million, including $275 million from private investors and $60 million from UK and U.S. governments. In China, a "2+N" industrial framework is emerging. The two national-level fusion enterprises—Fusion New Energy (Hefei) and China Fusion Energy (Shanghai)—have received nearly 30 billion RMB in funding. Meanwhile, numerous startups are exploring diverse routes: New York Energy and StarRing Fusion (spherical tokamaks), Energy Singularity (conventional tokamaks), and Hanhai Fusion (field-reversed configuration). Recently, StarRing Fusion was named one of MIT Technology Review’s "50 Smartest Companies" for its pioneering work in small-scale, economically viable fusion. Founded just four years ago and having raised tens of millions in two rounds, the company stands out for its innovative engineering approach. StarRing Fusion evolved from a清华大学 (Tsinghua University) research project—the Spherical Tokamak for Innovation and Science (SUNIST). Its core innovation lies in a novel, repeatable magnetic reconnection fusion operation scheme. Using high-temperature superconducting magnets, the system heats plasma internally through magnetic reconnection, eliminating the need for bulky external heating systems. The device is compact—only 8 meters in diameter and 10 meters tall—dramatically reducing construction costs and timelines. Instead of continuous operation, StarRing adopts a pulsed, engine-like cycle: heat plasma to fusion temperature in short bursts (about 6 seconds), let it react, then cool down before restarting. This avoids the challenge of maintaining stable high-temperature confinement for long durations. “We aim to make fusion engineering feasible,” says Chen Rui, founder and CEO of StarRing Fusion. “Our goal is to build a fusion reactor that costs around 1 billion RMB—far less than the tens to hundreds of billions for conventional designs.” The company has already achieved significant milestones. Its small-scale prototype, SUNIST-2, was built in just 279 days by a team of 50 people, reaching 17 million degrees Celsius and rapidly increasing plasma current from 100 kA to 480 kA. It achieved preliminary validation of its repeatable reconnection scheme within 11 months—marking a world-leading performance for a spherical tokamak of its class. The team of over 170 people includes 140 engineers and 30+ PhDs, many from Tsinghua’s fusion lab. Scientists not only solve physics problems but also translate them into precise engineering tasks—spanning mechanical, electronic, control, and optical systems—ensuring seamless integration. Currently, StarRing is building its mid-scale test facility, StarRing-1, aimed at achieving sustained fusion conditions and full engineering validation. For fuel, the company uses deuterium-tritium—chosen for its proven feasibility and economic viability. While some companies explore hydrogen-boron fusion, Chen Rui emphasizes: “We focus on the most practical path first. Once the system is mature, we can explore more advanced fuels.” Chen’s vision: achieve mass-produced fusion power plants with total costs under 10 billion RMB within a decade. Despite the promise, fusion remains capital-intensive and long-term. To ensure sustainability, StarRing has diversified its revenue stream—selling superconducting magnets and nuclear electronics—earning millions in 2024. On the U.S.-China fusion race, Chen believes China has a real chance to lead. “The U.S. has strong technical foundations, but China is advancing fast. Success lies in solving the weakest link—finding the right engineering solution.” According to FIA’s 2025 report, commercial fusion power is expected to arrive between 2031 and 2035. With rapid progress in compact designs, AI integration, and private investment, the dream of harnessing the power of the stars may finally be within reach.