First Trust Cloud Computing ETF
First Trust
Tracks the ISE CTA Cloud Computing Index — a modified equal-dollar-weighted basket (~65 holdings) spanning Infrastructure-, Platform-, and Software-as-a-Service, so it mixes hyperscalers, networking, and pure-play cloud names.
Top holdings
Cloud/AI data-center networking, top holding
Cloud infra for developers/SMBs
Google Cloud
Hybrid cloud / hyperconverged infrastructure
Hybrid cloud / Red Hat
AWS, the largest hyperscaler
Performance
July 2026: +6.27% (134.56 on 6/30/26 -> 143.00 on 7/31/26) — SKYY GAINED in July 2026 while SOX fell -20.6% and SMH -17.6% — the rout was AI hardware, not cloud software. Path: 2026 low 109.36 (3/31), melt-up to 144.57 (5/29, +31.9% off the Feb close), -6.9% June, +6.27% July including a +3.2% two-session bounce off the 7/29 low of 138.56. Computed 3Y annualized ~+21.1% (80.47 on 7/31/23), unreconciled to any provider table.
Recent moves
No notable rebalance/flow event surfaced; roughly flat-to-slightly-negative in 2026 after a strong prior three years.
Our take
The broadest cloud basket — by spanning IaaS/PaaS/SaaS and equal-weighting hyperscalers alongside small pure-plays, SKYY softened the 2026 software drawdown that hammered IGV/WCLD. The trade-off is dilution: its biggest holdings (Arista, DigitalOcean, Alphabet) pull in different directions, so it is a diversified-cloud hold rather than a sharp thematic bet.
The load-bearing fact marketing does not lead with: weights come from a binary checkbox, not a revenue measure.
A company deriving 3% of sales from infrastructure scores the same IaaS "1" as a pure-play. That single choice explains the holdings table better than any thematic story — as of 7/17/2026 Nutanix and Arista sit at 4.10% each while Microsoft sits at 3.39% and Amazon and Alphabet at 3.54% each, and IBM (2.77%) lands near DigitalOcean (2.96%). Note the cap is not what does this: no name is at 4.5%, and Microsoft earns the maximum Cloud Score of 6 yet still ranks below two mid-caps. It is theme-score bucketing plus drift since rebalance that flattens the hyperscalers. The honest description is not "the cloud": it is an equal-ish weight mid-cap enterprise-tech sleeve that includes the hyperscalers rather than being driven by them.
Thesis
SKYY sells "cloud" but structurally delivers something narrower: a mid-cap-tilted enterprise-tech basket whose weights are set by a trade association's binary yes/no checkboxes, not by how much cloud revenue a company actually earns. At 0.60% (issuer, as of 2/2/2026) against ~$2.80B AUM as of 7/17/2026, the fee buys a category definition that has decayed since the 2011 launch — by 2026 nearly every enterprise software firm is cloud-delivered, so the theme no longer draws a boundary cheap broad-tech doesn't already cross.
Tracks the ISE CTA Cloud Computing Index, 63 holdings ex-cash (issuer; 65 per provider). The Consumer Technology Association assigns each company a binary 0/1 for IaaS, PaaS and SaaS; weight follows a Cloud Score of 3xIaaS + 2xPaaS + 1xSaaS, with a 4.5% per-name backstop cap. Infrastructure is deliberately weighted 3x software. Not leveraged, no daily reset, no derivatives — volatility decay is not a mechanic here.
Assessment
- Breadth across IaaS/PaaS/SaaS (63 holdings ex-cash per the issuer), not the software-only slice — which cushioned the 2026 software drawdown vs pure-SaaS baskets.
- Theme-score bucketing with a 4.5% backstop produces real mid-cap exposure a cap-weighted tech fund gives you almost none of.
- Unlevered, physically replicated, 15-year live record (inception 7/5/2011) — no path-dependency or roll mechanics to model.
- Explicit 3x infrastructure tilt is a defensible view if you believe cloud economics accrue to the compute layer.
- Binary CTA scoring ignores magnitude — a token cloud segment earns the same score as a pure-play, so weights do not track actual cloud exposure.
- 0.60% for a definition, not a capability: the 2011 "cloud vs not-cloud" distinction has largely dissolved into "enterprise tech" by 2026.
- Structurally underweights the hyperscalers that won cloud — Microsoft scores the maximum 6 yet sits at 3.39%, below two mid-caps, because score bucketing flattens rather than ranks by economics.
- Top 10 is 33.78% of assets despite 63 holdings; diversification is shallower than the name count suggests.
- High overlap with tech allocations most holders already own, making the marginal exposure smaller than the 0.60% implies.
- 2026-07July 2026 ran the other way: the rout was hardware (SOX -20.6%, SMH -17.6%) while SKYY gained 6.27%. Breadth into infrastructure was the exposure at risk that month, not the cushion.
- 2026-07The compute-layer tilt was tested in July 2026 and lost: NBIS -31.1%, CRWV -27.9%, MRVL -37.0%, ALAB -35.6%. The gain sat in the accelerator duopoly (NVDA +0.3%, AVGO +3.1%), which score bucketing does not overweight.
Record
Verified: 1-year total return 10.70% and since-inception average annual 13.86% (stockanalysis.com, 7/21/2026), against ~-1.5% YTD in the source context. Attribution matters more than the level. Most long-run compounding came from mega-caps the score-bucketing scheme flattens to ~3.4-3.5%, not the pure-plays it structurally favors — the weighting has been a headwind to its own best ideas in a sector that proved winner-take-most. The same scheme is why 2026 has been comparatively mild: with no name above ~4.1%, the software de-rating that hit concentrated SaaS baskets hit SKYY diluted. Read the record as one mechanism producing both outcomes: flattened weights cost you in melt-ups and pay you in drawdowns. Trailing 3/5/10-year figures omitted — providers disagreed materially and none reconciled to the issuer's since-inception number.
- 2026-072026 was not a mild year after all: SKYY ran +31.9% from the 27 Feb close of 109.65 to 144.57 on 29 May, and finished July +9.9% YTD at 143.00. The flattened weights did not blunt that melt-up.
Risks & fit
- Classification risk: a CTA reclassification can force turnover and reshuffle weights for reasons unrelated to any company's fundamentals.
- Mid-cap high-multiple software concentration carries duration-like sensitivity to real rates; a rate shock hits this basket harder than cap-weighted tech.
- Theme obsolescence — if "cloud" ceases to be a distinguishable category, the index becomes an expensive, arbitrary tech sample.
- 0.60% compounds: at a 10% assumed gross rate over a decade it consumes ~8.7% of cumulative gains (~5% of terminal value), before any tracking difference.
- Rebalance mechanics reset winners back toward their score bucket, selling strength in a sector historically driven by a few extreme winners.
The structural case rests on one claim: cloud value keeps accruing to the hyperscalers the score bucketing flattens. It breaks if margin and growth migrate up-stack — if the mid-cap pure-plays (Nutanix, MongoDB, Cloudflare, DigitalOcean) out-compound Microsoft, Amazon and Alphabet over a multi-year window. In that regime the flattened weighting stops being a drag and becomes the point. Watch relative revenue growth and gross margin between the flattened mega-caps and the top-weighted pure-plays across several rebalances, not one quarter.
Suits an allocator who has already decided they want deliberate mid-cap enterprise-tech exposure with hyperscaler participation flattened rather than dominant, and who accepts a third-party trade body defining the universe. Poorly matched to someone seeking concentrated exposure to cloud's largest operators — cheap broad tech delivers those at a fraction of the fee — or to anyone treating it as diversification alongside an existing large-cap tech sleeve, where overlap is substantial.
0.60% expense ratio (issuer, as of 2/2/2026), roughly 6-7x a broad cap-weighted tech index fund and above several thematic-tech peers. The premium pays for index construction and category definition, not active security selection.