Three ways to own the same cash flow. Only one of them pays you 8%.
By Chris Berg · July 31, 2026
The five things
- Building beats buying by roughly 268 basis points. Public Storage is targeting an 8% stabilised yield on its $692M pipeline. The private market clears around 5.3%. That gap is the entire argument for being a developer in this cycle.
- Storage has the widest development spread of the major property types. Roughly 50 bps wider than industrial and between 1.7x and 2.5x the apartment spread — and every one of those numbers comes from the operators' own mouths.
- Storage is the only one of the three sectors where public and private have converged. Industrial REITs trade 3.4% below private value, apartments 10.2% below. Storage trades at a 0.8% premium. The public market is not telling you storage is mispriced; it is telling you the return is in the build.
- CubeSmart's balance sheet is the tell. Zero acquisitions, $42.5M of buyback at $38.96, and 15 stores sold into a Heitman JV at $197M. Management bought its own NOI at roughly 5.7% rather than third-party dirt at roughly 5.2%.
- What eats the spread is cost, not cap rates. Construction costs are running +5.2% year over year, and all three sectors are financing at negative leverage. The 268 bps is real, but it is being consumed from the cost side, not the yield side.
01The framework
Three doors, one cash flow
Every dollar aimed at self-storage has exactly three ways in. You can buy the shares of a company that already owns the real estate. You can buy the building from whoever owns it now. Or you can build it.
These are not three strategies. They are three prices for the same underlying stream of rent, and the only question that matters is which one is cheapest today. Almost nobody prints them on the same page, because doing it requires an equity analyst, a broker and a developer to agree on a definition of NOI. Here they are.
02Door 1
What the public market charges for storage NOI
No research house has published a current implied cap rate for these three companies — Green Street's is paywalled and the most recent free figure I could find is a UBS number from December 2025, which predates the National Storage Affiliates merger entirely. So I computed them from the filings.
| At the July 30, 2026 close | Public Storage | Extra Space | CubeSmart |
|---|---|---|---|
| Market capitalisation | $59,306M | $32,752M | $9,395M |
| Preferred equity | $4,731M | $48M | — |
| Net debt | $9,920M | $12,951M | $3,462M |
| Enterprise value | $73,957M | $45,752M | $12,857M |
| 2026E property NOI | $3,495M | $2,024M | $696M |
| 2026E ancillary NOI | $225M | $435M | $40M |
| Implied cap rate | 5.03% | 5.37% | 5.73% |
| property NOI only | 4.73% | 4.42% | 5.42% |
| on NSA/Canada run-rate | 5.36% | — | — |
The spread between the cheapest and dearest storage NOI in the public market is about 70 basis points — CubeSmart at 5.73%, Public Storage at 5.03%. That is not a rounding error. It is the market saying it will pay materially more for Public Storage's cash flow than CubeSmart's, and section 5 is about what CubeSmart's management decided to do about that.
03Door 2
What the private market charges
Both of the acquisitive REITs gave us the number on their calls this week. Extra Space's president Noah Springer put A-through-C market cap rates at "the high fours to the high fives." Public Storage bought or contracted for over $455 million year to date at low-5s yields, roughly 70% off-market — and it paid a loaded mid-5% going-in cap for National Storage Affiliates at about $180 per square foot.
CenterSquare puts the private market at 5.32% as of June 30. Newmark's tiering, reported secondhand, runs Class A at 5.05%, B at 5.95% and C at 6.75%.
So Door 2 sits somewhere between 4.75% and 6.75% depending on asset quality, with institutional product clearing in the low 5s. Which means Doors 1 and 2 are, for practical purposes, the same price. You can buy storage NOI through the stock market or through a broker and pay within a handful of basis points either way.
04Door 3
What it yields to build it yourself
Public Storage is the only public operator that put a number on this. On the July 30 call: $692 million across 47 projects, 4.0 million square feet, $432 million still to fund over 18 to 24 months, targeting an 8% stabilised yield.
CubeSmart has one ground-up project — a $28 million joint venture in New York opening in Q3 2027 — and disclosed no yield. Extra Space disclosed no development yield at all. So the entire public-market read on storage development economics rests on one sentence from one company on one call. That is a thin foundation and I want to be explicit about it.
Third-party sources put developer targets at 8–10%+, a 150–300 bps premium over acquisition cap rates — consistent with Public Storage's number, but from an aggregator rather than a primary source.
Why the spread exists at all
A cap rate prices existing cash flow. A yield on cost prices cash flow that does not exist yet. The 268 bps is the market's payment for three things you absorb: the 18-to-24 months before the building earns anything, the lease-up risk to stabilisation, and the possibility that costs run over. It is not free money. It is compensation for work and risk — which is precisely why it belongs to operators rather than allocators.
05The tell
CubeSmart bought none of its own asset class and $42.5 million of itself
The single most informative disclosure of the quarter is what CubeSmart didn't do. No acquisitions in Q2, none year to date. Instead: 1.1 million shares repurchased for $42.5 million at an average of $38.96, and an agreement to contribute 15 stores to a Heitman joint venture at a $197 million agreed value — roughly $219 per square foot — retaining 20%.
Read that as a capital allocation decision and it is unambiguous. CubeSmart's own NOI was available at roughly 5.7%. Third-party Class A dirt was clearing around 5.0% to 5.3%. Management bought the cheaper of the two, sold assets at private-market pricing to fund it, and told you so in the release: the JV "unlocks value from our portfolio and provides an accretive source of capital to support share repurchases."
06Cross-sector
Storage has the widest development spread of the major property types
The same three doors exist in industrial and apartments, and the operators disclose enough to price them.
| Sector | Public REIT | Private market | Build | Build over private |
|---|---|---|---|---|
| Self-storage | 5.27% | 5.32% | 8.00% | +268 bps |
| Industrial | 5.19% | 5.01% | 7.20% | +219 bps |
| Apartments — AvalonBay | 5.79% | 5.20% | 6.75% | +155 bps |
| Apartments — MAA | 5.79% | 5.20% | 6.25% | +105 bps |
| 10-year Treasury | 4.67% | — | — | — |
Two observations an allocator should take from that table. First, the risk-free rate is 4.67% — so buying stabilised storage through either of the first two doors earns you roughly 60 to 65 basis points over a Treasury, before any leverage, management, or capital expenditure. That is a thin premium for owning buildings. Second, industrial is the honest competitor: Prologis's 7.2% is an achieved yield on actual starts, not a target, and 74.7% of those starts are build-to-suit, which means much of the lease-up risk is already retired. Storage's 268 bps carries more risk than industrial's 219.
07What eats the spread
The threats are on the cost side, not the yield side
Construction costs are rising about 5% a year. Turner's Building Cost Index rose 1.4% in the quarter and 5.2% year over year; Mortenson has it at 6.8%; JLL expects acceleration in the second half. Climate-controlled single-storey storage runs roughly $80–120 per square foot in hard cost, with soft costs and FF&E adding 25–40% on top. A 5% cost increase on a project underwritten at an 8% yield on cost takes it to roughly 7.6% — which is 40 bps of the spread gone in twelve months without the cap rate moving at all.
Every one of these sectors is financing at negative leverage. CRED iQ's analysis of $26.1 billion of 2026 securitised originations puts self-storage cap rates 35 bps below the average mortgage coupon, industrial 30 bps below, multifamily 19 bps below. Debt does not improve the return on stabilised product right now; it dilutes it. That is another reason the development spread matters more than usual — it is the only leg where the yield clears the cost of capital by a comfortable margin.
And the supply that makes storage attractive is the same supply you'd be adding to. Forecast deliveries fall from about 54 million square feet this year to 45 million in 2027 and 38.6 million in 2028. The spread exists partly because so few people are building. A wave of capital chasing 268 bps is the mechanism by which 268 bps becomes 150.
08What would change my mind
Four things that would break this
- Public Storage's 8% turns out to be a stretch target. The entire development leg rests on one sentence. Prologis discloses achieved yields on stabilisations (6.3%) alongside targets on starts (7.2%) — a 90 bps gap between what gets promised and what gets delivered. If storage has a similar gap, the true spread is closer to 170 bps and the argument weakens considerably.
- Construction costs accelerate past 8%. Tariffs on Canadian imports take effect August 19, steel and aluminium are taxed at 50%, and JLL already flags H2 acceleration. Two more years at 6–7% closes the gap on its own.
- The private market cap rate widens. If sellers capitulate and Door 2 moves from 5.3% to 6.5%, buying stabilised product becomes competitive with building it and the case for development risk evaporates.
- Storage's public premium to private value inverts. Storage is the only one of the three sectors trading above private value. If it moves to industrial's 3.4% discount or apartments' 10.2%, then buying the shares becomes the cheapest door and the answer changes entirely.
09Methodology
How the numbers were built
Door 1 is computed, not cited. Enterprise value equals market capitalisation plus preferred equity plus net debt, at the July 30, 2026 close. Balance sheet inputs come from each company's Q2 2026 filing: Public Storage's 10-Q and supplement page 26, Extra Space's supplement page 17, CubeSmart's earnings release. Forward NOI comes from each company's own 2026 guidance reconciliation — PSA supplement page 27, EXR supplement page 15 — except CubeSmart, which publishes no supplement and no guidance NOI, so its NOI is annualised from the Q2 same-store table with a derived allowance for the 39 consolidated stores outside the same-store pool.
The convention matters and is arguable. I have used NOI before G&A and before capital expenditure, and I show the figure both with and without ancillary income (tenant reinsurance and third-party management fees). Including ancillary raises the cap rate because those businesses are not real estate; excluding it lowers the cap rate because enterprise value still contains their value. The truth is between the two columns and I have shown both rather than picking one.
Public Storage is the hardest to state cleanly. The NSA merger closed July 22 — after the balance sheet date but before the share price I am using. So the equity value reflects the merger and the NOI only reflects the partial-year contribution management guided to. I have shown both the as-guided figure and a run-rate figure that annualises the strategic acquisitions NOI, and the honest answer is somewhere in between.
Doors 2 and 3 are cited, not computed, and section 10 lists exactly how thin some of that sourcing is.
10What I did not verify
Required disclosure
- Door 3 rests on a single sentence. Public Storage's 8% target appears on the July 30 call only — not in the press release, not in the supplement. Extra Space and CubeSmart disclose no development yield at all. One data point is not a market.
- Every sector-level cap rate comes from one house. CenterSquare is the only source publishing Q2 2026 sector implied and private cap rates. No cross-check exists. A competing source, 2nd Market Capital, puts the all-REIT implied cap at 5.89% against CenterSquare's 5.39% and private real estate at 4.44% against CenterSquare's 5.98% — a material disagreement on both legs that I could not reconcile.
- The apartment development yields are stale. AvalonBay's 6.5–7.0% is from a February 2026 deck; MAA's 6.0–6.5% from a June 1 presentation. Neither restated a yield in its Q2 release, and Camden disclosed none at all. AvalonBay has also suspended guidance pending its merger with Equity Residential, which distorts any apartment series going forward.
- The cross-sector spread table does not exist anywhere as a published product. Every spread in it is either one operator's stated target or my arithmetic against a cap rate from a different source with a different methodology. This is the weakest methodological link in the piece.
- CubeSmart's non-same-store NOI is derived. The company publishes no supplemental — its 8-K contains the press release and nothing else — so consolidated property NOI for the 39 stores outside the same-store pool is my estimate, and its implied cap rate carries that uncertainty.
- Public Storage's post-merger preferred balance is partly derived. $4,350M at June 30 is hard, from the supplement. The additional ~$381M for the NSA Series T and U preferred assumes $25 par and is my computation.
- No cap rate was disclosed on the Heitman JV. The $197M agreed value on 15 stores is real; what it implies about pricing is not knowable from disclosure.
- Construction cost per square foot is secondary-sourced throughout. RSMeans and Cumming are paywalled; the $80–120 climate-controlled figure comes from a contractor site citing them, and a second source gives $65–85 for the same product. The two disagree by up to $35 a foot.
- CubeSmart's earnings call had not happened when this was written. The company released after the close on July 30; the call is July 31 at 11:00 a.m. Eastern. Everything here is from the release. Any management commentary — including the Heitman cap rate and the move-in rate trend — postdates this piece.
- CubeSmart does not disclose move-in or move-out rates, so the rate roll-down comparison that explained the PSA and EXR gap last quarter cannot be extended to it. I have not estimated it.
- No valuation conclusion. This piece prices three doors. It does not recommend one, does not model returns, and is not investment advice.