Commonwealth Fusion Systems

Chamath Palihapitiya

@chamath

What I Read This Week... 1) On Monday, JPMorgan Chase announced a sweeping plan to deploy $1.5 trillion in financing over the next decade toward what it calls “security and resiliency”. The categories are a blend of industrial strategy, national defense, and infrastructure modernization. “It has become painfully clear that the United States has allowed itself to become too reliant on unreliable sources of critical minerals, products and manufacturing – all of which are essential for our national security… It also needs to remove obstacles that stand in the way: excessive regulations, bureaucratic delay, partisan gridlock, and an education system not aligned to the skills we need.” - Jamie Dimon. JPMorganChase plans to focus on the following four key areas: 1. Supply Chain and Advanced Manufacturing: Critical minerals, pharmaceutical precursors and robotics 2. Defense and Aerospace: Defense technology, autonomous systems, drones, next-gen connectivity and secure communications 3. Energy Independence and Resilience: Battery storage, grid resilience and distributed energy 4. Frontier and Strategic Technologies: AI, cybersecurity and quantum computing Dimon concluded, “Hopefully, once again, as America has in the past, we will all come together to address these immense challenges. We need to act now.” 2) The Dutch government’s decision to seize control of Nexperia, a semiconductor firm owned by China’s Wingtech, marks a watershed moment in Europe’s industrial policy. The official reason was “recent and acute signals of serious governance shortcomings and actions within Nexperia” under former CEO Zhang Xuezheng. This move hopes to safeguard crucial technological knowledge and capabilities that could pose a risk to Dutch and European economic security. This intervention is rippling through the European corporate supply chain as China continues to tighten rare earth exports while the U.S. responds with further tariffs. 3) Google DeepMind’s latest research marks a profound turning point for AI. On Wednesday, Sundar Pichai announced that their model generated a novel hypothesis about cancer cellular behavior that was validated by scientists in living cells. The task they gave was for the model to find a drug that acts as a conditional amplifier. The model generated an unexpected insight into how tumors evade immune detection by suggesting a way to transform “cold” (invisible to the body’s immune system) tumors, which resisted treatment, into “hot” ones that respond to immunotherapy. The hypothesis was then tested in the lab and remarkably resulted in a roughly 50% increase in antigen presentation, which would make the tumor more visible to the immune system. “We can finally begin to simulate how real human cells behave—in context, in silico. This is where AI stops being just an analysis tool and starts becoming a model system for biology itself.” - David van Dijk, PhD, Associate Professor of Medicine, Yale If results continue to hold, this represents one of the first instances where an AI system not only mapped biological correlations but helped discover new biology. DeepMind also just announced a new partnership with fusion startup Commonwealth Fusion Systems, using AI models to accelerate plasma simulation and reactor optimization. If they succeed this will make SPARC the first magnetic fusion machine in history to generate net fusion energy.

Aakash Gupta

@aakashgupta

$15 billion is pouring into fusion energy right now and most people have no idea how close this actually is. WEST just sustained 50 million °C plasma for 1,337 seconds using 2 megawatts of heating power. That’s a duration record. But ITER, the $25+ billion megaproject being built on the same site in southern France, needs 50 megawatts of heating power to produce 500 megawatts of fusion output. WEST ran at 4% of ITER’s required input power. Duration at low power and duration at fusion-relevant power are completely different engineering problems. The real story is what’s happening around ITER while it bleeds money and time. Originally scheduled to produce first plasma in 2016, then 2020, then 2025, ITER has now pushed to 2034 for research operations and 2039 for actual deuterium-tritium fusion. The project has blown past $25 billion in costs with another $5 billion in overruns confirmed. Thirty-five countries, three decades of planning, and the reactor hasn’t turned on once. Meanwhile, private fusion capital has exploded from $1.7 billion in 2020 to over $15 billion by late 2025. Commonwealth Fusion Systems alone has raised nearly $3 billion, with Google signing a 200 MW power purchase agreement. Helion started construction on an actual power plant in Washington state, with Microsoft as its first contracted customer for 50 MW by 2028. Pacific Fusion emerged from stealth with $900 million. China is spending an estimated $3 billion per year on fusion and aims to have a plant online by 2031. The 22-minute record matters, but look at what it reveals about competitive dynamics. WEST is a testbed for ITER. ITER won’t run fusion experiments until 2039. By then, Commonwealth Fusion expects to have a commercial plant on the grid. Helion expects to have multiple plants deployed. China could have an operational facility for eight years already. Government-funded tokamak research proved the physics. Private capital is now racing to prove the economics. The duration record tells you the science is maturing. The $15 billion in private investment tells you the market has decided it can’t wait for ITER to figure out project management. The real race is who connects a fusion plant to a power grid first. And that answer probably won’t come from a 35-nation committee.

Dr Singularity

@Dr_Singularity

Wow. This is pretty huge news Google DeepMind has partnered with Commonwealth Fusion Systems (CFS) to use AI to accelerate the development of fusion energy, the same process that powers the sun. DeepMind’s AI will transform the way fusion is built and controlled. With its TORAX simulator, it can run millions of virtual fusion experiments before SPARC even turns on, finding the most efficient paths to stable, high output plasma. "When running at full power, SPARC will release immense heat concentrated onto a small area that must be carefully managed to protect the solid materials closest to the plasma." "In the initial phase of our collaboration, we’re investigating how reinforcement learning agents can learn to dynamically control plasma to distribute this heat" This collaboration is the beginning of a new era. AI guided fusion. It will give us clean, safe, and effectively limitless energy and make us post abundance tier civilization, and thanks to AI it can happen in the very near future.

Atoms Not Bits

@AtomsNotBits

BREAKING: Commonwealth Fusion Systems has secured $1 billion in funding, the largest private nuclear fusion round since 2021. The company is targeting completion of its first grid-connected fusion power plant by the mid-2030s.

Kyle Chassé 🐸

@Kylechasse

For 70 years, nuclear fusion has been 20 years away. It's not anymore. Helion's Polaris prototype just hit 150 million degrees Celsius, the highest plasma temperature ever recorded by a private company. The threshold for a commercially viable fusion machine is 100 million degrees. They're 50% past it. Commonwealth Fusion Systems sustained a reaction at 100 million degrees for 8.2 seconds. That's a 340% improvement over last year. Their reactor switches on in Cambridge in 2027. Helion is already building its commercial plant in Washington state. Signed contract to sell electricity to Microsoft by 2028. Power has always been the real power.

ApoStructura

@ApoStructura

Commonwealth Fusion Systems just raises another $1 billion, bringing the total to $4b They’re planning to reach net energy production in 2027. https://t.co/NGOHY7TCYF

Former Moderna CEO; joined CFS Board of Directors as an independent board member in 2026.