A boutique issuer from New York has just broken the ETF industry's growth record. The better question is why BlackRock, Vanguard and State Street didn't get there first?
On 2 April 2026, Roundhill Investments launched a small, narrowly focused fund called the Roundhill Memory ETF, ticker DRAM. It holds a concentrated basket of memory chipmakers, the companies producing the DRAM, High-Bandwidth Memory and NAND flash that AI data centres consume in vast quantities. Just two years earlier, Roundhill had 14 ETFs holding $958 million in assets.
What followed has no precedent. According to Bloomberg Intelligence, DRAM reached $6.5bn faster than any ETF before it, in just 36 days. It passed $10bn in roughly 30 trading sessions, the quickest asset-gathering pace ever recorded for an ETF, and crossed $20 billion in 54 days. Assets have since passed $25 billion, overtaking the iShares MSCI South Korea ETF (EWY), long the default route for US investors into SK hynix and Samsung.
Market returns did part of the work. By the $20bn mark, the fund had delivered a total return of nearly 180%. But the flows were real and they held up. Even during the sharp summer correction, DRAM took in $10.4 billion between the market's peak and trough. By any measure, this is one of the most successful product launches in the history of the industry. and what's more it came from a firm most European wealth managers had barely heard of.
Two traders and a thesis
Will Hershey and Tim Maloney co-founded Roundhill in late 2018. Both were Vanderbilt economics graduates of the class of 2011 and both CFA charter holders. Their paths had diverged after university. Hershey started at Yorkville ETF Advisors, later becoming Head Trader for Yorkville's long/short energy hedge fund. Maloney began as an FX and emerging markets salesperson at Morgan Stanley, then traded investment grade bonds for a $35 billion portfolio at Wells Capital.
Their thesis was blunt. Looking back, Hershey described the traditional ETF industry as "slow, backward-looking, and lacked innovation." Roundhill would build the opposite: fast, precise, first-to-market products for a generation of self-directed investors trading on apps rather than through advisers. The early catalogue reflected that ambition, with funds on sports betting, video games, cannabis and the metaverse. It also reflected the risk. By 2023, most of its products had been launched into fast-growing sectors that had since stalled.
The European misadventure
Roundhill's first attempt to go international ended badly. In 2022 it listed the Roundhill Ball Metaverse UCITS ETF on Xetra through a white-label platform, at almost the exact moment the metaverse theme peaked. The fund gathered just €1.7m in 18 months before Roundhill announced its closure and its exit from Europe. ETF Stream
Maloney's explanation at the time was candid. He said the firm could not find a sustainable growth model, despite strong support from its partners, and would refocus on the US. The whole theme struggled in Europe, too. Even the largest metaverse ETF there, from iShares, had amassed only $19.1m.
The lessons appear to have been absorbed. The episode showed that a white-label listing is not a distribution strategy. Theme timing matters more in Europe, where buyers are slower and more intermediated. And a product without proven US demand has little chance of pulling European money. This time the order of events is reversed: Roundhill is proving the product in the US first and only then looking abroad. In May, ETF Stream reported that Roundhill was anticipating strong European demand for DRAM, and in June that it was assessing appetite for a European-listed memory ETF. ETF Stream
Europe also presents a structural obstacle. DRAM's three largest holdings, SK Hynix, Micron and Samsung, make up almost 75% of the fund, a weighting that would breach UCITS concentration rules. Any UCITS version will have to be index-based and capped, which is a different product from the US fund. A competitor has also moved first. Defiance ETFs has listed the Defiance Memory UCITS ETF, also under the ticker DRAM, on Borsa Italiana and Xetra via HANetf's white-label platform, at a TER of 0.69%. A Roundhill return to Europe will now be a contest rather than a first mover's victory lap.
The CEO who came from the establishment
The turning point in Roundhill's story may be a hire rather than a product. Dave Mazza joined as Chief Strategy Officer in February 2023, and became CEO in April 2024 at the age of 41, with both founders keeping active roles in product development. THE ORG
Mazza's credentials are about as solid as the ETF industry offers. He began his career at State Street Global Advisors in research, portfolio management and product development, including leading the strategy and research team for SPDR ETFs, before moving into leadership roles at OppenheimerFunds and Rafferty Asset Management, the adviser behind Direxion. At Direxion he was Managing Director and Head of Product, running new product research and development, product strategy and strategic partner relationships. Across those roles he oversaw product strategy for more than 150 investment products with combined assets above $400 billion.
His arrival brought an important combination to the firm: incumbent-grade discipline in product governance and capital markets, applied to a challenger's appetite for risk. His stated strategy on taking over was to diversify beyond equity thematics into other areas that interest today's investors and traders. That meant 0DTE covered-call income funds, WeeklyPay single-stock products, leveraged vehicles and tax-efficient core funds. Those lines built the revenue base and distribution relationships that were ready when DRAM took off. He also supplied its one-line pitch: investors, he said, were realising that memory chips are the biggest bottleneck in the AI build-out. CNBC
What 0.65% on $25 billion buys
DRAM's gross expense ratio is 0.65% a year. On $25 billion, that is roughly $160 million in annual gross revenue from a single fund, before swap costs, sub-advisory fees and other expenses. For context, the whole firm managed under $1 billion two years ago, which would have generated perhaps $6–9 million a year in fees. Roundhill now reports more than $38 billion in assets under management, and one product has changed the firm's economics by an order of magnitude.
That money can pay for several things. The most obvious is launches: a semiconductor capital equipment fund covering lithography, deposition, etch, metrology and packaging is already on the list, alongside funds on silicon wafers, power semiconductors, photonics, neocloud providers and electronic components. Each one targets a candidate for the next bottleneck in the AI hardware stack. It can also fund a proper international distribution team instead of a listing and a hope. It can pay for senior hires: the firm has recruited Drew Pettit, after more than ten years in sell-side research, as Chief Investment Strategist. And it gives Roundhill strategic options, with a valuation that makes it both a plausible acquirer and a highly attractive acquisition target.
The risk is concentration. A large share of the firm's revenue now depends on one historically cyclical sector. DRAM fell more than 43% from its 2026 high during the summer shakeout. The central management task now is converting a windfall into a durable franchise before the memory cycle turns.
The innovator's dilemma, ETF edition
So why did the giants miss it? The uncomfortable answer is that, mostly, they didn't miss it. Their structure made a product like DRAM a bad bet for them before launch.
A boutique can run dozens of niche funds on the expectation that one breakout will pay for the rest, let's call that portfolio economics. For a flagship brand, a failed thematic fund costs more than its seed capital. It brings a closure notice, questions from governance committees and damage to the brand. The metaverse graveyard of 2021–23 is recent enough that no incumbent product committee has forgotten it.
Getting a pure-play oligopoly into a US fund required active management and total return swaps. Roundhill says it uses the swaps to stay compliant with RIC diversification tests. Incumbent product standards favour physically replicated, index-based funds with capped weights. A capped version would dilute exactly the exposure investors wants.
No memory index existed to license. Active ETFs under Rule 6c-11 let a small issuer define its own eligibility screen and launch within weeks while a theme is hot. An incumbent needs an index provider, a governance cycle and a slot in the launch calendar, so it seems speed to market matters.
Roundhill's buyers are self-directed, options-enabled retail traders. Incumbents sell to advisers, model portfolios and platforms whose due diligence teams would struggle to approve a 15-stock, swap-based fund with 40% drawdowns. Is this one of the best examples that shows that the retail market is now setting the trends in the ETF space?
They already owned part of the flow.
Memory stocks were already large weights in SMH, SOXX and EWY, so some of DRAM's inflows came out of incumbent products. That loss hurts less than money going straight from an incumbent to a competitor, and it blunts the urgency to respond.
The case against the incumbents is still real. In core beta, a fast follower can win by cutting fees. In thematic funds, liquidity reinforces itself. DRAM now has deep trading volume, an active options market and a 2x leveraged product built on it with T-REX. A later entrant charging 35 basis points would start with wide spreads and no options market. Waiting for a theme to be proven works in core products and loses in niche ones.
Europe is where the incumbents' passivity is hardest to defend. UCITS concentration rules force an index-based, capped design anyway, which is exactly the kind of product the large issuers build best, in their home market, for their adviser clients. Yet the first European memory ETF came from Defiance, a US boutique working through a white-label platform. iShares, Xtrackers and Amundi did not launch one.
Roundhill's success is not just a story about one lucky theme. It shows that the industry's biggest firms have organised themselves to manage what already exists, while the next generation of flows gets built by firms willing to be wrong in public many times over. DRAM has shown how much those misses can cost the incumbents.
Until next time.
Allan Lane