Pacer Data & Infrastructure Real Estate ETF
Pacer ETFs
Tracks a developed-market index of REITs earning ≥85% from data/infrastructure real estate — a pure-REIT vehicle of data centers and communication towers (no semiconductors, unlike DTCR).
- Concentrated — Top disclosed positions are 50%+ of the disclosed book.
Top holdings
Hyperscale data-center REIT
Largest data-center REIT
Communications towers
Records storage + data centers
European towers
US/intl towers
Recent moves
Same AI-datacenter tailwind benefits its DLR/EQIX core, but the heavy tower-REIT weighting (AMT/CCI/SBAC) muted returns vs compute-heavy DTCR; AUM drifted down (~$405M→~$397M over May-June 2026), suggesting net outflows.
Our take
The purest, lowest-tech datacenter-landlord play — concentrated in DLR/EQIX/AMT with no silicon, so it's the conservative way to own the physical AI real estate; the catch is that tower REITs have lagged, leaving it well behind DTCR on the same thesis.
The name oversells the AI angle. It reads as a barbell -- a hot data-center sleeve (EQIX/DLR ~31% combined, re-rating on hyperscaler leasing) welded to a roughly equal tower/telecom sleeve (~37-40%: AMT, SBAC, CCI, Cellnex) pressured by US carrier consolidation, flat domestic leasing and high rates.
The tower drag is much of why it lagged: the datacenter tailwind is diluted by a tower half of similar weight. It is also concentrated -- the top three (EQIX/DLR/AMT) are ~46% and the top six ~60% -- so you pay 0.49% for exposure largely replicable by holding EQIX/DLR/AMT directly. As a REIT it carries rate duration: at ~4% Treasury yields the complex de-rates, towers hardest. The clean read: SRVR is the conservative, no-silicon way to own data real estate -- but conservative has meant lagging, because the part of the AI-real-estate trade that worked (compute-adjacent silicon) is exactly what this fund excludes by design.
Thesis
SRVR's logic: own the physical real estate under the digital economy as a pure REIT, screening for developed-market landlords earning >=85% of revenue from data-center and communications-infrastructure property (Solactive GPR Data & Infrastructure index). The pitch is REIT-grade stability plus contracted, inflation-linked rents riding the same AI/data-buildout demand that lifts the compute names -- without owning any silicon or operating risk.
Passive, cap-weighted-with-caps REIT basket of ~64 names, but top-heavy: EQIX + DLR + AMT are each >14% (~46% combined), trailed by a mid-tail of Iron Mountain, SBAC, Crown Castle and Cellnex (~4% each). No leverage, no swaps, no semiconductors. 0.49% fee; ~2.9% dividend yield.
Assessment
- Pure real assets: contracted, often inflation-linked leases from investment-grade tenants (hyperscalers, carriers) -- genuinely lower operating risk than owning the chip cycle
- No leverage and no swap/collateral machinery -- returns are the underlying REITs, with none of the daily-reset decay or counterparty layering of leveraged AI vehicles
- Owns the irreplaceable EQIX/DLR data-center oligopoly, whose pricing power and record leasing are real beneficiaries of AI capacity demand
- ~2.9% dividend yield gives a cash-return component that the growth-tilted AI-compute ETFs lack
- ~37-40% of the portfolio is tower/telecom REITs (AMT, SBAC, CCI, Cellnex) -- a structurally slow-growth, rate-pressured sleeve roughly matched in weight by the data-center names, offsetting much of their gains
- Top-heaviness: EQIX/DLR/AMT are ~46% of assets (top six ~60%), so this is effectively a 3-stock bet dressed as a 64-name index
- 0.49% fee for a concentrated basket largely replicable with a handful of direct REIT holdings at ~zero cost, or via VNQ at 0.13%
- Rate-sensitive REIT duration -- at ~4% Treasury yields the fund de-rates regardless of AI demand, and towers de-rate most
- AUM has drifted down (~$405M to ~$366M) on net outflows, signalling investors are rotating toward compute-inclusive vehicles
Record
The record undercuts the AI-real-estate narrative. Since its May-2018 inception SRVR has compounded only ~4.8%/yr over eight years -- a mediocre absolute result, and the 2026 leg has turned outright negative: -5.7% trailing 1y, -13.8% over three months, -10.4% in the last month (as of Jul 21, 2026), with a max drawdown of ~41% and ~21% annualized volatility. What drove it: the data-center sleeve (EQIX/DLR) did re-rate on record hyperscaler leasing, but the tower/telecom half was flat-to-down on carrier consolidation and rate pressure, and the whole REIT complex carries duration that hurt as yields held near 4%. This is precisely why the hybrid DTCR -- which adds a semiconductor/memory sleeve (~20-25%, approx; SK Hynix, Micron, Marvell) -- crushed SRVR on the identical 'AI real estate' thesis: the part that worked was the silicon SRVR deliberately excludes. SRVR's conservatism was real, and so was its cost in returns.
Risks & fit
- Rate/duration: sustained ~4%+ long yields de-rate all REITs, towers most, independent of AI demand
- Tower-REIT structural drag: US carrier consolidation and flat domestic leasing cap AMT/CCI growth
- Concentration: top three (EQIX/DLR/AMT) ~46% means one landlord's leasing miss or guidance cut swings the fund
- Thesis dilution: excludes the compute/silicon exposure that has driven the AI-real-estate outperformance
- Small and shrinking AUM (~$366M, net outflows) raises spread/liquidity and viability questions vs larger peers
The critique flips if the tower sleeve stops being dead weight and rates ease: a decisive fall in long-term Treasury yields would re-rate the whole REIT complex, and re-accelerating tower leasing (carrier capex, edge/5G, tower M&A) would let AMT/CCI add rather than offset. If, over a 2-3 year window, data-center rent growth compounds fast enough that EQIX/DLR carry the fund past a compute-inclusive peer like DTCR on a risk-adjusted basis, then 'conservative = lagging' no longer holds and the pure-landlord design is vindicated.
Maps to an allocator wanting real-asset, dividend-paying REIT exposure to digital-infrastructure landlords with explicitly no semiconductor or operating-company risk, accepting REIT rate-duration and a large tower-REIT weighting as the price of that purity. It maps poorly to anyone whose actual goal is the AI-compute buildout -- that exposure lives in the datacenter-plus-silicon hybrids and semi/power funds, not a pure-landlord REIT screen.
0.49% is steep for a concentrated passive REIT basket -- ~4x Vanguard's broad-REIT VNQ (0.13%). With EQIX/DLR/AMT ~46% of assets, much of the exposure is replicable near-zero cost; the fee buys tail diversification, not hard-to-access assets.