This is about legendary short-seller Jim Chanos betting against data center REITs (real estate companies that own data centers). He argues that companies like Digital Realty (DLR) and Equinix (EQIX) are overvalued: they spend $11 to make $1 in revenue, burn billions in cash yearly, and face rising interest rates. Their biggest tenants, like Amazon and Microsoft, are also building their own data centers, becoming competitors. Chanos says if you like the industry, just buy Amazon or Microsoft instead.
Wall Street legendary short seller Jim Chanos, in an episode of Business Breakdowns, articulated his bearish thesis on U.S. data center REITs. The core argument is that since 2016, hyperscalers have been reshaping the industry landscape, leading to structural challenges for traditional data center R
Jim Chanos, the legendary short seller on Wall Street, articulated his bearish view on U.S. data center REITs in the Business Breakdowns program. The core argument is that since 2016, hyperscalers have been reshaping the industry landscape, leading to structural challenges for traditional data center REITs. Chanos points out that these REITs face severe cash burn issues, their asset values are difficult to assess, and high leverage risks are amplified in a high-interest-rate environment. He questions their profit margins and depreciation models, arguing that investors underestimate the impact of a macro tech slowdown. Additionally, he expresses concerns about the broader commercial real estate market. Chanos advises management to respond rationally to short sellers and avoid common mistakes.
Chanos categorizes enterprise data storage into three models: on-site self-management, co-location (the core business of traditional data center REITs), and cloud services (hyperscalers such as AWS, Azure, and Google Cloud). He points out that the co-location model faces structural decline, as cloud service providers have significantly increased capital expenditures since 2016, steadily eroding market share.
Key data support:
Chanos emphasizes: "If your incremental return on invested capital is negative, then it doesn't matter how fast you're going. In fact, if you grow, you're liquidating faster."
Chanos exposes a critical accounting issue: these REITs claim that maintenance capital expenditures account for only 10% of total capital spending, implying an assumed asset useful life of 150 years ("150 years is, of course, absurd").
Mechanism breakdown: According to insider explanations, if replacing an HVAC system can be claimed to "attract new tenants or increase rent," the entire expenditure is classified as growth capital spending rather than maintenance. Chanos considers this an "accounting joke."
Data chain: These companies' capital expenditures are typically 150-175% of depreciation and amortization. Excluding growth factors, Chanos argues that depreciation is not only a real expense but may even be underestimated.
Chanos points out that DLR's free cash flow after dividends and before asset sales is negative $2.7 billion (approximately $230 million per month), relative to its $30 billion market cap — a staggering cash burn rate.
Leverage Data:
Valuation Contradiction: Chanos notes that these REITs are still trading at 100x P/E. He argues that data centers are essentially technology operating businesses, not pure real estate, and should be valued by tech company standards — and a 100x P/E is unsustainable even among tech stocks.
Chanos observes that in 2020-2021, private equity (PE) acquired data centers at 25-30x EBITDA, with Digital Bridge even buying Switch at 40x EBITDA. However, in a rising interest rate environment, these deals are becoming "regrettable."
Key Signals:
Chanos calculates that if DLR's capitalization rate rises from 5.4% to 8%, the stock would face a downside of over 70%.
Chanos points out that hyperscalers are the largest tenants of traditional data center REITs, yet they are also building their own data centers at lower costs. This dynamic—where "your largest competitors are also your largest tenants"—is highly unfavorable for landlords.
Competitive data:
Chanos acknowledges that a slowdown in tech spending could affect both supply and demand: on one hand, it would reduce demand for traditional data centers; on the other, it might also slow hyperscalers' new construction investments. However, he argues that the ultimate outcome remains uncertain and requires observation of whether traditional data centers can "regain market share."
Falsification condition: Chanos explicitly states that if these companies' ROIC begins to exceed their cost of capital, he would change his view. But the current gap is too wide, and he believes "from here to there is going to be a pretty bumpy road."
Chanos recalls his firm's history of shorting commercial real estate in the late 1980s, noting similarities between the current environment and that period. He specifically points out that office rents and occupancy peaked in 2018 (before the pandemic), and the post-pandemic trend of remote work has further weakened demand.
Key Mechanisms:
1. Cap rates do not deduct corporate management fees: Management fees for publicly traded REITs typically range from 50 to 150 basis points, but many companies are shifting operating expenses to SG&A, making NOI appear higher
2. Leasing costs are capitalized: Broker commissions, tenant improvements, rent-free periods, and other incentives are capitalized, and these costs actually rise in a contracting environment
3. External regulatory risks: Taking New York City's Local Law 97 as an example, it requires buildings over 25,000 square feet to achieve full green energy compliance starting next year, and even the most modern buildings (such as SL Green's One Vanderbilt) are non-compliant — this is essentially a "disguised increase in property taxes"
Chanos warns: "If you were paying a four or five cap for an office, the real economics might have been a one or two cap. That just doesn't work if your duration of your leases is seven to 12 years."
Chanos cites Netflix’s Reed Hastings as the "gold standard"—Hastings rebutted the short report point by point, without emotion or personal attacks on the short seller. He also recalls that when he was a young sell-side analyst, a company CEO invited him to tour the operations and spent an entire day rationally discussing their differences.
Common mistakes:
Chanos emphasizes that short sellers should not be held to a higher standard than longs: "I keep saying, well, you should see the 48 buy recommendations I get in my portfolio every morning in my inbox. No one says boo about that. But when a short seller issues a report, they are held to a higher standard—that’s the industry we chose."
Chanos argues that to generate excess returns in the market, one must hold a view that differs from the consensus ("variant perception") and be willing to bet on it. One advantage of short selling is that it is easy to hear opposing views (since longs will actively attack), which helps test assumptions.
Core principle: If, after thorough testing, you still believe your factual judgment and viewpoint are correct, then you have a potentially profitable variant perception.
| Position | Analyst Stance | Key Data |
|---|---|---|
| Digital Realty (DLR) | Strong Short | Incremental ROIC 2%; Annual cash burn $2.7B; Leverage 9x; Implied cap rate 5.4% vs. private market 8-10% |
| Equinix (EQIX) | Short | Incremental ROIC 5-6%; Implied cap rate 5.6%; Interest cost <3% |
| Switch | Short (Acquired by Digital Bridge at 40x EBITDA) | Transaction valuation overpriced |
| Six Tera | Short | Trading at 9% cap rate; Potentially a key customer of DLR |
| GDS Holdings | Short | Trading at 9% cap rate |
| DCRU REIT (DLR Singapore subsidiary) | Short | Trading at 9% cap rate; DLR holds >20% stake |
| Amazon (AWS) | Long (Indirectly via long QQQ) | Cheaper, better, and higher return than DLR |
| Microsoft (Azure) | Long (Indirectly via long QQQ) | Same as above |
| Google (Cloud) | Long (Indirectly via long QQQ) | Same as above |
1. "When incremental return on capital is negative, the faster you grow, the faster you liquidate" (Chanos) — DLR requires $11 in capital for every $1 of additional revenue, generating $0.50 in gross cash flow, with a payback period exceeding 20 years.
2. "Your biggest competitor is also your largest tenant" (Chanos) — Hyperscalers are both the largest customers of traditional data center REITs and are building cheaper data centers themselves. This conflict of interest structure is extremely unfavorable for landlords.
3. "Maintenance capital expenditures are misclassified as growth expenditures" (Chanos) — Replacing an air conditioning system, if claimed to "attract new tenants," is counted as growth capex, resulting in an implied asset life of 150 years — an "accounting joke."
4. "If you like the industry, buy Microsoft or Amazon — they are cheaper and have better businesses" (Chanos) — Hyperscalers have returns, growth rates, and return on capital that are multiples of traditional data centers, yet their cash flow multiples are lower — the "ultimate irony."
5. "The private market has shifted from buyer to seller" (Chanos) — In 2020-2021, PE firms acquired data centers at 25-40x EBITDA; now they are selling at 8-10% capitalization rates, signaling a valuation system under reconstruction.
6. "Don't just look at surface-level NOI" (Chanos) — The true economic return on commercial real estate may be 300-400 basis points lower than the surface cap rate, as management fees, capitalized leasing costs, and regulatory risks (e.g., New York Local Law 97) are all hidden.
7. "Office rents and occupancy peaked in 2018" (Chanos) — This predates the pandemic; post-pandemic remote work merely accelerated the trend, rather than being its root cause.
8. "The best way for management to respond to a short report is with a calm, fact-based point-by-point rebuttal" (Chanos) — Reed Hastings sets the gold standard; emotional attacks only attract more short sellers.