You think you bought an aerospace and defense fund. You did not. The Global X Defense Tech ETF holds 50 stocks, and not one of them is Boeing or GE Aerospace. Those two names anchor every large US defense benchmark, and this fund's rulebook throws both out.
What it owns instead is a portfolio whose biggest position is a software company. Palantir sits at 11.73% of the fund as of Sept. 29, ahead of RTX at 9.08% and General Dynamics at 8.32%. In the rival iShares fund, GE Aerospace alone is roughly a fifth of the book and Boeing is another large slice. SHLD holds neither.
Here is the part almost nobody has priced. The index caps any single company at 8%, and Palantir is nearly four points above that line. The cap is not enforced continuously; it is enforced twice a year, and the next reset takes effect Dec. 1, 2026. The fund will sell a chunk of its best performer on a calendar, not on a view.
That is the fund in one line: a concentration limit that looks like protection and behaves like a scheduled trade. What follows is what you own, how the machinery works, what it costs, what it has returned, and the strongest case against it.
What you actually own
Start with the Sept. 29, 2026 snapshot. The five largest positions were Palantir at 11.73%, RTX at 9.08%, General Dynamics at 8.32%, Lockheed Martin at 8.25% and Northrop Grumman at 7.81%. Together they are 45.19% of the fund, and the top two alone are 20.81%.
Below them sit the foreign primes and the second tier of contractors. BAE Systems is 4.65%, Hanwha Aerospace 3.83%, Rheinmetall 3.69%, Thales 3.66%, L3Harris 3.66%, Saab 3.60%, Leonardo 3.56% and Elbit Systems 3.06%. Ten names carry 64.68% of the money, fifteen carry 79.42%, and the top twenty carry 86.98%. The other 30 positions split 13.02% between them.
That tail is where the fund gets strange. Planet Labs is 0.71% of the book, Karman 0.48%, Voyager 0.25%, BigBear.ai 0.20%, Droneshield 0.16%, York Space 0.15%, Red Cat 0.14%, Kraken Robotics 0.14% and BlackSky 0.11%. Several of these lose money at the operating line, and each is too small to matter on its own.
Then there is the split that surprises people. Roughly 36.66% of the fund is not American: Britain 6.25%, South Korea 6.13%, Germany 5.23%, France 4.74%, Italy 3.55%, Sweden 3.53%, Israel 2.99% and Norway 2.07%. That is currency risk as much as country risk, and the holdings file carries euro, won, shekel and lira balances beside the shares.
The winners and losers tell the same story. AeroVironment, a 0.86% position, is down roughly 39% this year after a June 2026 restatement and a disclosed material weakness. Kratos, at 1.26%, is down roughly 37% and trades above 270 times trailing earnings on thin margins. Neither moves the fund, and both are exactly what the pure-play screen exists to admit.
What does 45% in five names mean on the day those five fall together? It means the fund is those names, because nothing else in the book is large enough to cushion the blow. Watch Palantir and RTX, which at 20.81% combined are the whole fund in miniature.
How the sausage is made
SHLD is an index fund, but not in the way the phrase implies. It tracks the Global X Defense Tech Index, which Global X owns and writes, while the daily calculation is done by Mirae Asset Global Index in Gurgaon, India. Both sit inside the same corporate family as the fund's adviser, which bought Global X in 2018.
The selection rule is a revenue test, and it is strict. A company qualifies only if at least half its revenue comes from cybersecurity, defense technology, or advanced military systems and hardware. Defense technology means AI, geospatial and ISR work; hardware means robots, drones, weapons, space launch, satellites and radar.
Two exclusions do most of the work. Civilian firearm makers are out, and so is any company whose revenue is mostly commercial aircraft components. That is why Boeing and GE Aerospace fail the screen rather than merely being underweighted. Eligibility runs across developed and emerging markets, excluding China, India, Kuwait, Pakistan, Russia and Saudi Arabia.
The plumbing has floors worth knowing. New names need $200 million of market value and existing ones $160 million. Only the most liquid share class of any company qualifies, and the final step takes the 50 biggest survivors by market value.
Weighting is the part that matters most. Names are weighted by free float, the shares actually available to trade rather than the whole company. An 8% cap applies to any single name. There is also an aggregate rule: once the companies above 5% add up to more than 40% of the fund, the rest get clipped to 4.5%.
Reconstitution happens twice a year. Selection day is the first Friday of May and November. Weights are frozen five index business days before the rebalance and take effect on the next index business day. The most recent reset, scheduled from the close of May 2026, took effect June 1, and the next takes effect Dec. 1, 2026.
The wrapper does the rest. The fund runs full physical replication, buying the actual shares rather than sampling the index. Shares trade on NYSE Arca all day, so you get a quoted price whenever you want instead of a once-daily print. Creations and redemptions happen in kind, and that is the quiet tax advantage: a departing institution takes stock rather than a check.
Two details are easy to miss. The fund may lend up to a third of its assets to borrowers, which adds a little revenue and a little counterparty risk. It also owns locally listed foreign shares, so real currency positions sit inside the portfolio. Data sites that report 56 holdings are counting cash, four currency balances and a payables line; the fund holds 50 equities. Watch the Dec. 1 weight reset, because that is where these rules stop being theory.
The bill
The fund charges 0.50% a year, with no fee waiver, no temporary reduction and no expiry date hiding a subsidy. Of the three big US defense ETFs, this one is the most expensive: ITA charges 0.37% and XAR charges 0.35%. SHLD is about a third more expensive than ITA and roughly 43% more than XAR.
Put the fee in human terms. A $10,000 position pays about $51 in the first year, per the prospectus fee table: half a percent of the balance, plus a year of assumed growth. Ten years of that is roughly $500 if the money never grows, and closer to $630 if it compounds at 5%.
The costs people forget never appear on a fact sheet. The 30-day median bid-ask spread is 0.05%, so a round trip leaks a few basis points to the market maker. The fund traded about 0.02% above its net asset value on Sept. 29, which is tight and tells you the creation mechanism works. Neither is the real drag.
The real drag is the gap to its own index. The fund trailed its benchmark by 0.93 percentage points in fiscal 2025, by 1.02 points in calendar 2025, and by 0.84 points a year since inception. That is more than the fee on its own. Turnover of 32.79% adds a tax layer, because the December trim of the winners is a realized sale unless redemptions absorb it.
Watch the tracking number as closely as the fee. A thematic fund that gives up roughly a point a year to a rulebook its own parent writes is expensive at any headline price.
The scoreboard
Through Sept. 29, measured on closing prices, the picture is ugly short-term and remarkable long-term. SHLD is down roughly 6.4% this year while the S&P 500 is up roughly 13%. Over one year the fund is down roughly 12% and the broad index is up roughly 16%.
Over three years the arithmetic flips. SHLD has returned roughly 36% a year, against roughly 26% for ITA and roughly 23% for the S&P 500. There is no five-year record to compare, because the fund only began in September 2023. Since then it has compounded at roughly 35% a year, and every point of that was earned before this year began.
The rival matters, because it is the fund most buyers think they are getting. ITA is down roughly 2.3% this year and up roughly 2% over one year. That means SHLD has given up roughly four points to the plain aerospace and defense index fund this year alone. The gap is not manager error; it is the missing Boeing and the missing GE.
The fund's own benchmark comparison is cleaner and less flattering. In fiscal 2025 the fund returned 60.10% against the index's 61.03%, and calendar 2025 was 74.55% against 75.57%. Same methodology, same two portfolio managers from day one, and the wrapper costs roughly a point a year.
Assets tell the same story with a delay. The fund held $4.83 billion at the end of November 2025, $7.24 billion at the end of August 2026 and $6.56 billion on Sept. 29. Roughly $2.4 billion of new money arrived in 2026, most of it after the best returns were booked.
Valuation and yield round out the case. On Sept. 29 data the fund trades at roughly 29 times trailing earnings, against roughly 25 times for the S&P 500 as measured by SPY on Sept. 30 data. Its forward multiple of 23.5 implies roughly 22% earnings growth, and its 0.6% yield is about a third less income than the broad index pays near 1.0%.
The drawdown is the number to sit with. The fund is roughly 23% below its 52-week high, which is what a 45% top-five weight looks like when those five names re-rate together. Watch whether another year like this one leaves the three-year record intact.
What you're really betting on, and the bear case
Now argue the other side the way a short seller would. The bet is sovereign and allied defense budgets compounding into software, drones and munitions, with Europe and Korea re-arming for a decade. That is a real thesis, and it has already been paid for.
Start with momentum. The fund is down roughly 6.4% this year while the S&P 500 is up roughly 13%, a nineteen-point gap that says the re-rating already happened. The sector went from a 15% premium to the S&P 500 in February to a 12% discount by late September. In September alone, 42 of the 50 stocks in XAR fell.
Then look at what the fund is actually long. Palantir is 11.73% of the book and trades on a venture-style multiple, and if it halves that is roughly six percentage points off the fund. The forward multiple of 23.5 assumes roughly 22% earnings growth, and a company that misses that assumption does not drift, it re-rates.
The structure has a timing problem built in. On Dec. 1, 2026, the rules force the fund to cut its five biggest positions back toward the aggregate 40% cap. That is roughly five percentage points of the fund, or about $340 million at the September asset level, roughly $245 million of it Palantir. The fund sells winners and buys laggards on a fixed date.
Concentration is the other structural fact. The top ten names are 64.68% of the fund, and it is officially non-diversified, which means the tests that govern most funds do not apply. A mistake at RTX or a multiple reset at Palantir is not a drag on this fund. It is the fund.
A third of the book is foreign, which brings currency risk most US defense funds never carry, and the European primes have already re-rated hard. Part of the tail is weaker than volatile. BigBear.ai, Red Cat, BlackSky, Voyager and Planet Labs all lose money at the operating line, and AeroVironment carries a June 2026 restatement, a material weakness and a class action.
What would prove the bears right? Palantir breaking its 2026 low on a guidance miss while still the largest weight. A December reset into a falling tape, where the forced sale lands at the worst moment. A third quarter of net redemptions, which would push the fund to sell the same names the rebalance is already trimming. Any one of those turns a bad year into a broken thesis.
How to use it
Treat this as a satellite, not a core holding. It is a 50-stock, non-diversified, single-theme fund with 45% of its money in five names and a 23% drawdown already on the board. Size it so a repeat of this year does not change your plan.
It does not replace a plain aerospace and defense fund, and that is the sentence to remember. SHLD shares only about 14 holdings with ITA, and those names are more than 40% of its weight. It owns no Boeing, no GE Aerospace, and a third of it is foreign. If you want the US primes plus commercial aerospace, this is not that fund.
Who it suits: an investor who specifically wants defense software, drones and the European primes, and who reads the index rules before buying. Who it does not: anyone who wanted plain aerospace and defense. Also not for anyone who cannot tolerate a third of the book in foreign currency, or who would sell after a year like this one.
Weight the December reset before you buy anything. It tells you what the fund will own for the next six months, and it is the clearest test of whether the methodology is enforced as published.
The verdict
SHLD is a well-built wrapper around a real thesis that is not the thesis on the label. It is a defense-tech fund with a software company on top and a rules-based sell order dated December.
What to watch next. The Dec. 1, 2026 reconstitution and the Palantir cut, which is the largest mechanical trade on the fund's calendar. The semi-annual review in May 2027, where the tail names face the same screen and several may not survive it. And whether the December reset lands into a stable tape or a falling one.
One line to finish. This is a concentrated bet sold under a diversified label, and the difference between it and the defense fund you thought you bought is named Boeing and GE Aerospace.
Analysis, not advice. We may own these names and we may sell them at any time.
