DRAM vs LUMA: Which Is the Better Pick in 2026?

Overview

DRAM and LUMA are both thematic technology ETFs betting on infrastructure buildouts for artificial intelligence, but they target different hardware layers. DRAM focuses on memory semiconductors — the chips that store and retrieve data in computing systems — while LUMA targets photonic and optical components, including transceivers, fiber cables, and lasers that move data between systems. Both are nascent funds launched in 2026 with no distributions, but they differ sharply in scale and selectivity.

How they differ

The biggest difference is fund size and maturity. DRAM has accumulated $26.0B in assets since its 04/02/2026 launch, while LUMA stands at just $9.71M since 07/14/2026. That difference reflects both timing and investor appetite: DRAM's memory-chip thesis has broader institutional recognition, while LUMA's photonics exposure is narrower and newer.

Second, DRAM's expense ratio of 0.65% undercuts LUMA's 1.00%, a meaningful gap for buy-and-hold investors, especially if the smaller fund struggles to scale.

Third, their underlying exposures sit on different parts of the AI infrastructure stack. DRAM holds memory semiconductors — essential to every data center, gaming rig, and edge device — making it a core-layer play. LUMA's optical and photonic components are higher-margin specialty hardware for data interconnection, concentrated among fewer public companies and relying partly on private-company exposure, which carries valuation opacity and liquidity risk.

Who each is best for

DRAM: Fits investors seeking broad exposure to the memory-semiconductor cycle through a single vehicle, with conviction that memory demand from AI infrastructure, cloud computing, and consumer electronics will drive sustained chip pricing and capacity expansion.

LUMA: Fits investors with a targeted thesis on optical data-movement infrastructure as a bottleneck in AI scaling, willing to accept smaller scale, higher expense drag, and exposure to illiquid private holdings in exchange for a more specialized hardware bet.

Key risks to know

• Concentration risk in semiconductor supply chain: DRAM's focus on memory chips ties it to a handful of dominant manufacturers and geographies; disruptions to fabrication capacity, trade restrictions on advanced chip sales, or oversupply cycles can swing valuations sharply. LUMA's narrower optical and photonics market compounds this risk, with even fewer publicly traded players and greater reliance on private-company valuations that may not reflect public-market price discovery.

• Early-fund liquidation and closure risk: Both funds are extremely young with limited track records. LUMA's $9.71M makes it vulnerable to asset hemorrhage if the photonics narrative falters or investor interest wanes; small, thematic ETFs have historically closed when assets fall below operational viability thresholds.

• Private-company valuation opacity in LUMA: By holding private companies alongside public ones, LUMA introduces mark-to-model risk — private holdings are revalued by the fund manager, not by public markets — and may not reflect current investor sentiment or liquidity conditions. This also complicates tax-loss harvesting and creates potential NAV gaps at the expense of shareholders.

• AI infrastructure demand assumptions: Both funds rest on the premise that AI scaling will require sustained, rising memory and optical bandwidth. If AI adoption plateaus, capex budgets shift, or efficiency gains reduce hardware demand per unit of computation, both portfolios face headwinds regardless of individual company execution.

Bottom line

DRAM offers larger scale, lower costs, and exposure to a more established semiconductor subsector; LUMA pursues a more specialized optical-infrastructure thesis with higher fees and no institutional AUM cushion. If you want broad memory-semiconductor exposure at a reasonable cost, DRAM's profile aligns better; if you believe optical bottlenecks will drive disproportionate returns and can tolerate illiquidity and a younger fund, LUMA may fit. Both funds lack performance history and distributions, so past returns cannot guide evaluation — thesis conviction and risk tolerance should drive the choice.