Global X Just Launched an SMR ETF. Its Own Holdings Show How Small the Industry Still Is.

On October 15, 2026, Global X launched a fund that lets anyone buy the small-modular-reactor industry in a single trade. The Global X Small Modular Reactor ETF, ticker SMRZ, holds roughly 15 companies that design, build, or fuel the reactors that are supposed to deliver cheap, factory-built nuclear power to a grid that is suddenly hungry for it. The pitch is compelling: data centers are devouring electricity, governments are approving reactor designs, and technology companies are pouring money directly into reactor developers.

What the fund's own disclosures reveal, though, is how much of that future still exists only as a pitch.

What's in the fund

SMRZ is actively managed, charges a 0.75 percent gross expense ratio (that is $75 a year on every $10,000 invested), and targets about 15 companies globally. To make the cut, a company generally must derive at least half its revenue from small modular reactors or the fuel and service agreements tied to them, or qualify as a development-stage company. Those criteria sound broad until you look at what actually qualifies.

The top 10 holdings, as of the close of business on October 14, 2026, account for 97.84 percent of the fund's weight. The top five, NuScale Power, X-Energy, Oklo, Nano Nuclear Energy, and Terrestrial Energy, account for 74.74 percent. The top three alone, NuScale, X-Energy, and Oklo, make up 53.37 percent. This is not a diversified sector portfolio. It is a handful of reactor startups alongside established industrial firms like GE Vernova, KEPCO Engineering and Construction, and BWX Technologies, which together carry 13.7 percent of the weight.

The most honest document in nuclear power right now may be a fund's holdings table: it lists exactly what exists, and the list is short.

The scenario that justifies it

The launch announcement leans on the International Energy Agency's January 2025 outlook on the path to a new era for nuclear energy, and specifically on the scenario in which global nuclear capacity more than doubles by 2050 with small modular reactors playing a central role. In that scenario, the world builds more than 1,000 SMRs by 2050, with a combined capacity of 120 gigawatts. Investment in SMRs rises from about $5 billion today to more than $25 billion in 2030, and reaches a cumulative $670 billion by 2050.

Those figures imply a very specific set of assumptions. One thousand units totaling 120 gigawatts means an average reactor of about 120 megawatts. Dividing $670 billion by 120 gigawatts gives roughly $5.6 per watt of installed capacity, the cost level the scenario assumes factories can hit. And reaching 1,000 units by 2050 means installing about 40 reactors a year, every year, for 25 years, starting from a base of essentially zero commercial units in operation.

None of that is impossible. But it is worth comparing with the historical pace of nuclear construction, where a decade to build a single large reactor is common, and where costs have routinely exceeded projections. The bet embedded in SMRZ is not that nuclear power works (it does). The bet is that factories, not construction sites, can make reactors cheap, and that regulators and customers will move fast enough to keep 40 reactors a year rolling off assembly lines.

What has actually happened

The announcement cites a genuine milestone: on September 29, 2026, the US Nuclear Regulatory Commission approved construction of the first commercial small modular reactor in the United States, at Tennessee's Clinch River site. That approval is real, and it matters. It is also one reactor. Around the same time, filings for other proposed units sat at very different stages of the regulatory process, including Holtec's phased construction permit application for two SMR-300 units, dubbed Pioneer Units 1 and 2, at the Palisades site in Michigan. Europe's accumulation of firsts continues as well, with a heat-only reactor design receiving its first public bank loan from the European Investment Bank.

Meanwhile, the demand side is moving faster than the supply side. Technology companies are investing directly in reactor developers, trying to secure the firm, round-the-clock power that variable renewables alone cannot guarantee for data centers. The launch materials make a fair point about engineering heritage, noting that the US Navy has operated small reactors since the USS Nautilus entered service in 1954. That is genuinely reassuring about the underlying physics. It says much less about whether commercial SMRs can be built for $5.6 a watt.

What would have to be true

Four conditions need to hold for the launch scenario to materialize. Licenses must arrive on schedule and on budget, a track record the industry does not yet have. Factory-built reactor costs must fall to the levels the scenario assumes, which can only be proven once factories are actually building. Government support, from loan guarantees to streamlined permitting, must stay in place across multiple election cycles. And customers, especially utilities and technology companies, must sign firm contracts before the first units have proven their economics.

None of these are reasons to dismiss the sector. The first commercial construction approval, the pipeline of NRC filings, the European project finance, and the direct tech investment all point in the same direction. But a fund that concentrates 97.84 percent of its weight in ten companies, most of which have never delivered a commercial reactor, is an honest map of the gap between the industry that exists and the one the scenario needs. That gap is either an opportunity or a warning, and the same document can read as both. Nothing here is investment advice: references to securities are not a recommendation to buy or sell.