The first fusion stock. The first data centre ban. A SpaceX startup targets Mars.

General Fusion began trading on Nasdaq as GFUZ on July 13 following its SPAC merger with Spring Valley Acquisition Corp. III, becoming the first publicly listed fusion energy company in the world — with the stock rising approximately 38–40% on debut to give the company a market cap exceeding $380 million.

The company entered public markets with approximately $150 million in cash — anchored by an oversubscribed $105–108 million PIPE — to advance its Magnetized Target Fusion (MTF) technology through a series of plasma milestones on its LM26 machine, targeting the Lawson criterion by 2028 without superconducting magnets or high-powered lasers.

The listing coincided with the Fusion Industry Association's record annual report: $4.48 billion invested in private fusion companies in the past 12 months, bringing the cumulative global total to $14.24 billion — a sector that barely existed as an investable asset class five years ago now has its first publicly traded pure-play company.

GFUZ is the first publicly accessible price signal for commercial fusion — a sector that has been exclusively private-market funded for two decades. Institutional investors who could not participate in fusion rounds now have a listed vehicle, and GFUZ's trading performance will set a public benchmark against which every other fusion company — CFS, Helion, TAE, Proxima — will be implicitly valued.

General Fusion's MTF approach is structurally distinct from the tokamak and stellarator designs pursued by CFS and Proxima: it mechanically compresses plasma using a lithium liner, avoiding the superconducting magnet infrastructure that defines its competitors' capital costs. The LM26 machine has already reached approximately 0.72 keV electron temperatures (~8.4 million °C) — a measurable, independently verifiable technical result, not a simulation.

The listing is TIME magazine's #1 GreenTech Company 2026 going public in the same week that global fusion funding hits a record — the confluence signals that fusion is transitioning from a research narrative to a capital markets storythat investors need to price, even before net energy gain has been demonstrated commercially.

GFUZ is now publicly accessible on Nasdaq — the first liquid entry point into pure-play fusion. However, the significant SPAC redemptions meant General Fusion's actual treasury is ~$150M, not the $230M+ originally anticipated. The Lawson programme's 2028 milestones are the near-term catalysts: 1 keV plasma heating, then 10 keV, then Lawson criterion. Each creates a specific binary event for the stock.

The pro-forma equity value at the merger announcement was approximately $1 billion versus the debut market cap of ~$380M — reflecting SPAC redemption dynamics rather than a fundamental downgrade of the fusion thesis. Investors who believe General Fusion reaches its 2028 milestones should assess current pricing against that original valuation reference.

Risk caveat: General Fusion has never generated commercial revenue and has no path to electricity production before 2030 at the earliest. GFUZ's value is entirely a function of technical milestone execution — and no fusion company has yet publicly demonstrated the Lawson criterion at commercially relevant scale. The $150M treasury funds the milestones, not a power plant.

TerraFirma raised $115 million on July 14 in a round led by Kleiner Perkins, with angels from SpaceX, Anduril, and Hadrian, to scale its semi-autonomous heavy construction equipment platform — retrofitting existing excavators, tractor loaders, and bulldozers with sensors, cameras, and remote control systems operated via Xbox controllers and a "click to dig" 3D interface.

One TerraFirma operator can currently supervise 3–5 machines simultaneously, moving more than 300% more material per shift than a conventional single-machine crew. The company will use the capital to hire approximately 300 employees, open a Texas manufacturing facility, and build a centralised mission control centre from which operators pilot fleets of machines across multiple job sites simultaneously.

TerraFirma is currently working on commercial projects across housing, energy, transportation, and manufacturing — including a sports arena in Spicewood, a Starbucks in North Austin, and a power substation in New Braunfels — as well as US government international infrastructure projects in challenging operating environments, a phrase that points toward classified or semi-classified construction in austere locations.

The construction industry faces a structural labour deficit: Associated Builders and Contractors estimates it needs 349,000 additional workers in 2026 just to maintain current output levels. TerraFirma's remote operation model — one skilled operator managing 3–5 machines — directly addresses the productivity side of that equation without requiring human bodies on the ground in dangerous or remote environments.

Marc's analytical frame from the CNBC story: TerraFirma is one of a growing cohort of SpaceX alumni companies applying aerospace engineering discipline to earthbound industries — alongside Castelion (hypersonic weapons) and Relativity Space (additive manufacturing for rockets). The SpaceX flywheel of talent compounds beyond launch: the same engineers who built iterative, rapid hardware systems at Hawthorne are now applying that methodology to construction, defence, and manufacturing.

The long-term thesis is infrastructure for interplanetary settlement: SpaceX's $86B IPO and NASA's push for permanent Moon and Mars bases have made construction-in-space a fundable vision rather than a science fiction premise. TerraFirma's goal of becoming "the largest construction company in the solar system" is unusual in a pitch deck — but as the first company founded specifically to build autonomous construction infrastructure for planetary expansion, it has no direct competitor.

TerraFirma is privately held at this stage, with no disclosed valuation. Kleiner Perkins and Bain Capital Ventures are the primary institutional holders; the next funding round — which will follow proof of scalable fleet operations and potentially additional government contract wins — is the next investable event.

The US government international infrastructure work signals potential defence and special operations adjacency that could provide non-cyclical, high-margin revenue alongside commercial construction. Anduril's angel participation is particularly notable: Anduril has a track record of helping SpaceX alumni companies navigate the transition from commercial to government contracting.

Risk caveat: autonomous heavy construction in variable, uncontrolled outdoor environments is meaningfully harder than autonomous driving — every job site is unique, unpredictable, and physically dangerous in different ways. TerraFirma's model keeps skilled human operators in the loop, but the 300%+ productivity claim is based on current operations at early commercial scale. Whether that advantage holds at the scale of large civil infrastructure projects has not yet been demonstrated.

NRC staff published a Safety Evaluation Report (SER) formally recommending that the NRC issue TVA a construction permit for a BWRX-300 SMR at Clinch River, Oak Ridge, Tennessee — the first-ever NRC staff SER recommending a commercial SMR construction permit in US history, with the mandatory uncontested hearing now scheduled for August 13, 2026, ahead of the original December timeline.

The SER confirmed TVA meets five formal criteria: compliance with the Atomic Energy Act and NRC regulations, conformance to applicable guidance, reasonable assurance of safety margin in the preliminary design, ability to construct in conformance with the permit, and demonstrated financial and technical qualifications to execute the project.

The BWRX-300 — a 300 MWe boiling water reactor designed by GE Vernova Hitachi Nuclear Energy — is a Generation III+ design that does not require active cooling systems and is designed for factory modular manufacturing, enabling a radically shorter on-site construction programme than conventional large reactors.

The SER is a required, legally binding deliverable in the NRC's commercial licensing process — not a preliminary opinion or informal signal. It is the formal document that triggers the hearing phase and puts NRC commissioners on a timeline to issue or deny the permit. The August 13 hearing date, ahead of the original December schedule, signals that the NRC is treating Clinch River as a priority review.

This is the civilian commercial nuclear regulatory path — not the DOE research pathway. While Issues 13 and 14 covered DOE zero-power criticality demonstrations under the Reactor Pilot Program, those are research milestones under DOE authorization. An NRC construction permit is what authorises a utility to pour concrete and physically construct a commercial nuclear plant that will generate electricity for the grid. The two programmes run in parallel, and both are progressing simultaneously.

The Clinch River project uses the same BWRX-300 design at the centre of the US-Japan-South Korea trilateral SMR agreement from Issue 14, and GE Vernova Hitachi is the common technology provider across multiple planned commercial deployments in the US, Canada, UK, and Poland. An NRC construction permit at Clinch River would become the global regulatory reference for the BWRX-300 design.

The primary listed beneficiaries are GE Vernova (NYSE: GEV) as BWRX-300 technology provider and turbine supplier, TVA (federally owned, no public shares), and BWX Technologies (NYSE: BWXT) as a key fuel and component manufacturer for advanced reactor programmes. An approved construction permit would trigger procurement contracts across GEV's nuclear supply chain.

The August 13 hearing and subsequent NRC commissioner vote is the near-term binary event for the Clinch River project and for BWRX-300 deployment globally. A positive commissioner vote would unlock construction spending and signal to other BWRX-300 applicants worldwide that the design has cleared the most rigorous regulatory scrutiny in the industry.

Risk caveat: the final construction permit has not yet been approved by NRC commissioners. The uncontested hearing on August 13 is a procedural step, not a rubber stamp — commissioners retain full authority to modify or deny the permit following the hearing. Cost and schedule risks from Vogtle Units 3 and 4 remain the reference case for how US large nuclear construction can deviate from plan.

Governor Hochul signed an executive order on July 15 imposing the nation's first statewide moratorium on new hyperscale data centres — pausing environmental permits for facilities with 50 MW or more of peak power for up to one year, directing the DPS to develop a GEIS and requiring data centres to either supply their own power or pay a premium to use the shared grid through the Energize NY proceeding.

The executive order is separate from and more targeted than the Responsible Data Center Development Act passed by the New York legislature — which set a 20 MW threshold — that Hochul has not yet signed. The EO uses a 50 MW threshold and takes effect immediately, with Hochul citing the need to protect electricity ratepayers whose bills have risen 68% since 2019, while the legislative bill continues under review.

Hochul is also pursuing repeal of New York's sales tax exemption for large data centres, which currently provides significant subsidy to hyperscale operators — a move that would require legislative action and represents a structural reversal of the incentives that have attracted data centre investment to the state.

Fourteen US state legislatures have introduced data centre restriction bills; New York is the first to actually enact a statewide pause. Maine's legislature passed a similar moratorium, but Governor Janet Mills vetoed it. Arizona enacted a three-year data centre sales tax exemption freeze but stopped short of a construction moratorium. New York's action sets a legal and political template that other states' governors can follow by executive order, without waiting for legislative approval.

The moratorium creates an immediate capital redirection signal: data centre developers with projects planned for New York will redirect investment to Virginia, Texas, Georgia, and Ohio — states that have explicitly positioned themselves as data centre-friendly. Real estate investment trusts with New York data centre exposure, and hyperscalers with New York capex commitments, face near-term project delays.

The 50 MW threshold is the key design choice: a facility below 50 MW — a conventional, non-hyperscale data centre — is not affected. The moratorium specifically targets AI-focused hyperscale infrastructure, effectively writing a policy distinction between general-purpose computing and AI compute infrastructure into New York law for the first time. That distinction has implications for how states regulate AI physical infrastructure as a category separate from technology generally.

The most direct investment signal is geographic reallocation of hyperscale data centre capex: Virginia (Dominion Energy, data centre REIT equities including Equinix and Digital Realty), Texas (ERCOT, Oncor Electric Delivery, and the Chevron/Microsoft Kilby model), and Ohio are the primary beneficiary states. Data centre developers and power utilities in those states are the clearest financial winners from New York's pause.

The Energize NY "bring your own power" requirement — even before the moratorium lifts — accelerates demand for co-located power solutions like Project Kilby (Chevron/Microsoft, Issue 12) and nuclear-powered data centres (Aalo/Deployable Energy, Issues 13-14). Any future New York data centre approval will need to come with power self-sufficiency built in.

Risk caveat: executive orders can be reversed, modified, or subjected to legal challenge by industry groups arguing economic harm. The AI Infrastructure Coalition has already warned the moratorium will cost New York economic activity and shift investment to competing states. Whether the GEIS process produces a workable framework within 12 months — or simply extends the moratorium — is the gating question for the entire project pipeline.

Ant Group led a 500M yuan ($73.58M) pre-Series A round in humanoid robotics company Zeroth — its 12th investment in the humanoid robot sector since early 2025 — spanning robot manufacturers (Galaxea, Unitree), component and supply chain specialists (Linkerbot, Hypershell), and AI software companies (Genrobot AI, AheadForm), assembling a full-stack humanoid ecosystem rather than betting on a single winner.

Alongside portfolio investments, Ant has built its own robotics subsidiary RobbyAnt, which unveiled the R1 humanoid robot in September 2025 targeting service environments — pharmaceutical sorting, medical consultations, guided tours — and is also developing an AI and robot-friendly version of Alipay to integrate payment infrastructure directly with robot services at scale.

Zeroth, the company in Ant's latest deal, has secured 30,000+ robot orders and reported operating revenue growth of 600% in the first half of 2026 versus the prior year; CEO Renjie Guo has stated plans to begin North America and Europe sales this autumn, once local compliance requirements are cleared.

Ant's 12-deal strategy is deliberately portfolio-wide, not winner-take-all: rather than concentrating on a single humanoid maker, Ant is building exposure to robot manufacturers, component suppliers, AI software, and embodied AI data infrastructure simultaneously — the same diversified approach Amazon uses in logistics robotics and Google uses in AI compute. The goal is ecosystem control, not product monopoly.

China's humanoid market has government backing that US competitors do not: the China Ministry of Industry and Information Technology has made humanoid robots a national technology priority, and investment by Ant, Baidu, Tencent, and Huawei is occurring simultaneously — a pattern that historically produces rapid industrial scaling rather than fragmentation. The MIIT's mandate for 10,000 humanoids in Chinese factories (covered in Issue 10) is the demand-side pull; Ant's portfolio is the supply-side assembly.

The Alipay integration angle is the long-term signal most Western investors are missing: Ant processes over $17 trillion in payments annually through Alipay. A humanoid robot that integrates natively with Alipay — for delivery, retail, service — creates a payments-plus-robotics flywheel that no Western humanoid company can replicate, because no Western competitor controls comparable payment infrastructure.

Ant Group is not publicly listed following the 2020 halting of its planned IPO. The primary listed exposure is Alibaba Group (NYSE: BABA / HKEx: 9988), which owns approximately 33% of Ant. Zeroth, Galaxea, Unitree, and other portfolio companies are all privately held.

The Zeroth investment is notable for its Horizon Robotics chip dependency — Zeroth's robots currently run on Horizon Robotics processors, a Chinese semiconductor company listed in Hong Kong (HKEx: 9660). Any Western regulatory restrictions on Chinese semiconductor exports would directly affect Zeroth's production capacity.

Risk caveat: Ant's robotics push is a strategic diversification bet by a company that was operationally restricted following the 2020 IPO halt and subsequent regulatory interventions. The portfolio has significant China-market concentration, and a Western investor audience should assess geopolitical screening carefully — both Ant Group and the MIIT context imply Chinese government alignment that requires flagging for any cross-border capital decisions.

Airhive acquired Carbyon on July 14, combining two complementary solid-sorbent DAC technologies: Airhive's fluidized-bed process (mineral sorbent particles suspended in fast-moving airflow, sub-$500/tonne cost demonstrated at its Storm-1 system in Alberta) and Carbyon's ultra-fast sorbent (90% saturation in ~100 seconds — approximately 200x faster than conventional DAC systems).

The combined company will co-develop UnionDAC — targeting 60,000 tonnes/year of CO₂ removal by 2032 at Teesside in the UK's northeast (phase 1: 20,000 t/year by 2030; phase 2: +40,000 t/year by 2032) — which would exceed the capacity of any existing DAC facility globally and connect to the Northern Endurance Partnership's subsea CO₂ storage infrastructure in the North Sea.

Airhive is simultaneously deploying on-site DAC at a Coca-Cola Europacific Partners bottling facility to produce low-carbon, food-grade CO₂ for beverages — a commercial application that demonstrates the DAC-to-utilisation revenue model that does not depend on voluntary carbon credit markets.

This is the first significant M&A transaction in the European DAC sector — and Bloomberg explicitly noted it occurs as "demand and policy support for DAC have become less predictable just as the industry is reaching commercialisation." M&A consolidation in an early-stage deep tech sector typically signals that a shakeout is beginning: the companies with the strongest technology, project pipelines, and investor backing survive; the rest struggle for capital.

The EU Carbon Removal and Carbon Farming Regulation (CRCF) has now formally certified DAC as a carbon removal methodology — one of the regulatory pre-conditions the sector has needed to unlock public procurement and compliance-market demand. The UK is planning to integrate DAC-based credits into its Emissions Trading Scheme by 2028–2029. These two policy signals create the offtake backdrop for the 60,000-tonne UnionDAC project to access government procurement revenue alongside voluntary credits.

The Coca-Cola utilisation partnership is strategically significant: it is the first example of food-grade CO₂ from DAC at a commercial beverage facility. CO₂ for food and beverage applications sells for substantially higher prices per tonne than carbon credit revenue ($200–$400/tonne equivalent for food-grade CO₂ vs $100–300/tonne for removal credits), and the buyer is contractually obligated to take delivery regardless of voluntary carbon market volatility.

Both Airhive and Carbyon are privately held; financial terms of the acquisition were not disclosed. Carbyon had raised approximately $17.9 million in total, including a €15.3 million Series A in 2024 backed by Lowercarbon Capital, ISAI, and Invest-NL. The combined company's next financing round — likely a significant Series B to fund the UnionDAC project — will be the primary external investment event.

The UnionDAC 60,000-tonne Teesside project creates an infrastructure-scale revenue anchor for the combined company that neither Airhive nor Carbyon could have secured independently. Once operational, it would also generate the long-term certified removal dataset that institutional credit buyers (compliance markets, corporate Scope 3 accounting) increasingly require before making large offtake commitments.

Risk caveat: DAC remains expensive — Airhive's Storm-1 sub-$500/tonne demonstration is still well above the $50–100/tonne at which DAC would be cost-competitive at gigaton scale. The Cascade technology is in development, not yet demonstrated. And "market oscillations" is CEO language for what Bloomberg described more plainly: policy and demand support for DAC has become less predictable, which is the structural headwind the combined company will need to navigate even with its stronger technology and project pipeline.

Architecture as a carrier of health — buildings designed from the ground up around the biology of the people inside them

Technology as an amplifier of ancient wisdom — biometric data and natural systems integrated at the design level

Wellbeing as a measurable asset — longevity science embedded in every structural and environmental decision

Community as a protective layer — a curated global ecosystem of health-conscious individuals

IP as scalable value — a multi-platform ecosystem for designing human life, built to grow with the science