A data center is a building purpose-built to house computing and storage: rows of server racks, the power to run them, and the cooling to keep them from melting. They are the physical backbone of cloud services, streaming, and now artificial intelligence, and demand for them has surged as AI workloads multiply. For investors, that demand story is magnetic. It is also a sector where the barriers, power, land, and capital, are so high that institutional money dominates and very few vehicles reach the typical accredited investor. We will be plain up front: Baker 1031 does not have a benchmark yield or a track record to show you for data centers. This is a brand-new sector for our firm, and a rare one for individual investors generally. This guide is built on general, well-established market context, framing data-center DSTs as an emerging and uncommon category, and explaining what an accredited investor should scrutinize if one ever crosses your desk. For the mechanics of the structures themselves, see our guides to the 1031 exchange and the Delaware Statutory Trust.
Key Takeaways
- Baker 1031 has no benchmark yield and no track record for data centers; this is an emerging sector for the firm and a rare one for individual investors, so lead with the specific deal, not sector hype.
- Power is the binding constraint. A facility's secured electricity and grid interconnection can matter more than its lease, and unsecured power is a risk that has nothing to do with tenants.
- Strong hyperscale credit does not solve technology obsolescence. Underwrite the capex plan and the re-leasing picture as hard as the current lease, because the building can age faster than the tenant.
What a data center is
Strip away the mystique and a data center is a building organized around three things: power coming in, computers running, and heat going out. The real estate value is not in the walls. It is in the electrical capacity, the redundant power and backup generation, the cooling systems, and the connectivity that let tenants run servers reliably around the clock. A data center is measured less in square feet than in megawatts of power it can deliver and dissipate.
That orientation makes data centers unlike almost any other property type. An office is valued on location and finishes; a data center is valued on how much power it can pull from the grid, how reliably it can keep that power and cooling running, and how well it connects to networks. The most important number in the building is often the one on the utility interconnection, not the one on the lease per square foot.
We tell anyone looking at this sector to think in terms of power and infrastructure first. The building is a shell around an electrical and mechanical system, and the economics, the demand, and the risks all flow from that system rather than from the conventional real-estate fundamentals investors are used to.
What is driving the demand
The demand surge is real and well documented at a general level. Cloud computing moved a huge share of business and consumer software off local machines and into remote facilities, which created steady, growing need for data-center capacity. Artificial intelligence has poured fuel on that. Training and running large AI models takes enormous, dense computing power, which translates directly into demand for more data centers, and for ones built to handle far higher power density than older facilities.
Behind that demand sit a small number of very large customers, the hyperscalers, companies like Amazon Web Services, Microsoft, Google, and Meta, that operate cloud platforms at global scale. Their capital spending on computing capacity has been substantial, and it shapes the whole sector. When the largest technology companies commit to long-term capacity, developers build to serve them, and the demand signal cascades down through the market.
We will keep the specifics general on purpose, because precise figures in this sector move fast and we will not put numbers in your hands that we cannot stand behind. The directional picture is what matters: digital and AI demand has driven strong absorption of data-center capacity, and the largest technology companies are the ones underwriting much of it. That is the tailwind everyone points to, and it is genuine. The constraint, which gets less airtime, is the more interesting part of the story.
Power is the binding constraint
Demand alone does not build a data center. Power does, and power is the hard limit. A modern facility needs a large, reliable electrical supply, and getting it means securing an interconnection to the grid, which utilities can only provide where capacity exists and where transmission can carry it. In many of the markets where tenants most want capacity, the grid is already stretched, and the wait to connect a new large load can be long.
This is why the sector's growth is constrained less by demand or even capital than by electricity. You can have a willing hyperscale tenant, the land, and the money, and still be unable to build because the local grid cannot deliver the megawatts on the timeline the tenant needs. Power availability, interconnection queues, and transmission capacity have become the real gatekeepers, and they explain why development has concentrated in markets with available power rather than simply where demand is highest.
For an investor, the lesson is that a data-center asset's value is tied to something most real estate never has to think about: the local electrical grid and the facility's secured power. A building with locked-in, ample power in a constrained market holds a genuine advantage. One that depends on grid capacity it has not yet secured carries a risk that has nothing to do with tenants or leases and everything to do with the utility.
In most property types, demand sets the pace. In data centers, the grid does. The binding constraint is power, and a facility's secured electricity can matter more than its lease.
Gerald F. "Jerry" Baker, IIIHyperscale, turnkey, and powered shell
Data centers are not a single product, and the differences shape both the risk and who the tenant is. Three broad models cover most of the market, and they sit on a spectrum from least to most finished by the developer.
A powered shell is a building delivered with the structure, the power capacity, and the connectivity in place, but with the tenant responsible for installing the computing and most of the internal systems. A turnkey facility is delivered fully built out and operational, ready for the tenant to move servers in. A hyperscale build is a large facility, often built to suit, leased to a single hyperscale tenant on a long-term lease, sometimes designed to that tenant's exact specifications. Each shifts the build-out cost and the risk between owner and tenant differently, and each attracts a different kind of occupant.
| Model | What the owner delivers | Typical tenant | Build-out risk |
|---|---|---|---|
| Powered shell | Structure, power, connectivity | Operators who fit out their own space | Mostly on the tenant |
| Turnkey | Fully built, operational facility | Enterprises wanting ready space | More on the owner |
| Hyperscale build-to-suit | Large facility to tenant's spec | A single hyperscale tenant | Shared, long lease |
General industry framing of data-center models, not Baker 1031 figures. These categories describe how build-out and risk are split between owner and tenant; specific deals vary widely.
The model matters because it determines who carries the heavy capital cost of fitting out the facility and who bears the technology risk inside it. A powered shell leased to an operator is a different investment than a turnkey facility the owner built and equipped, even if both sit in the same market.
Who signs the lease
At the top of the market, the tenants are some of the strongest credits in the economy. Hyperscale leases go to companies such as Amazon Web Services, Microsoft, Google, and Meta, large, well-capitalized technology firms whose ability to pay rent is not in serious doubt. A long lease to a tenant of that caliber is, on the credit alone, an attractive income stream, and it is much of why institutional investors have pursued the sector so aggressively.
Below the hyperscalers sit colocation and enterprise tenants, companies that rent space and power in shared facilities rather than building their own. Their credit varies, and a multi-tenant colocation facility carries a different risk profile than a single hyperscale lease, with more tenants to manage and re-lease but less dependence on any one of them. The right way to read a data-center deal, as with any income real estate, starts with who is on the lease and how long it runs.
There is a wrinkle specific to this sector, though. A strong tenant on a long lease still leaves the owner exposed to what happens to the building's technology over that term, and to whether the facility can be re-leased to a future tenant whose power and cooling needs may differ from today's. The credit answers one question. It does not answer the obsolescence question, which we turn to next.
Technology obsolescence and heavy capex
Data centers carry a risk most real estate does not: the technology inside them, and sometimes the building's design itself, can age out. Computing has trended toward higher power density, and AI workloads in particular demand far more power and cooling per rack than facilities were built for a decade ago. A data center designed for yesterday's density can become harder to lease to tenants whose needs have moved on, even if the structure is sound.
Keeping a facility competitive takes capital, and a lot of it. Power and cooling systems wear and need replacement, density requirements rise, and standards shift. The capital expenditure to maintain and upgrade a data center over a long hold can be substantial, and it is a recurring claim on the income the building produces. An investor reading only the headline yield, without accounting for the capex needed to keep the facility relevant, is reading half the picture.
This is where the strong-tenant story and the obsolescence story meet. A long hyperscale lease can insulate an owner during the lease term, but what the building is worth at the end, and whether it can attract the next tenant, depends on whether it kept pace with technology. We would underwrite the re-leasing and upgrade picture as carefully as the current lease, because in this sector the building can become outdated faster than the tenant becomes uncreditworthy.
Why so few data-center DSTs exist
Put the pieces together and you can see why this sector is dominated by institutions and why so few data-center DSTs reach individual investors. The barriers to entry are extreme. Securing large blocks of power and grid interconnection is hard and slow. Suitable land near that power is scarce. And the capital required to develop and equip a modern facility is enormous, far beyond a conventional commercial building. Each of those barriers, on its own, screens out most buyers. Together they leave the field to large, well-capitalized players.
That concentration of capability is exactly why data-center DSTs are rare. The DST structure is built for fractionalizing stabilized, income-producing real estate so that 1031 investors can buy a piece sized to their exchange. Data centers, with their high capital intensity, technology risk, and the deliberate management constraints a DST operates under, are a difficult fit for that model. Most data-center real estate is held by REITs and private institutional funds rather than offered through DSTs.
So an accredited investor encountering a data-center DST is encountering something uncommon, and that rarity is itself a reason for extra scrutiny. It is not a mature, widely available 1031 product the way net-lease retail is. We would want to understand very clearly why this particular asset was structured as a DST, who the sponsor is, and how the structure's constraints interact with a sector that can need active capital and management.
How a data center would sit inside a DST
If a data-center DST does come to market, the mechanics resemble any other. A Delaware Statutory Trust holds title to the property, a sponsor manages it, and each investor owns a beneficial interest sized to the exact dollar amount a 1031 exchange has to absorb. That precision is the same reason investors use DSTs in any sector: an exchange must hit a specific number, and a fractional interest can be sized to the dollar.
The friction is the structure's constraints meeting the sector's needs. Under Revenue Procedure 2004-86, the rules that let a DST interest qualify for 1031 treatment, the trust operates inside tight limits, the seven deadly sins: after closing the sponsor cannot raise new money, cannot refinance, and cannot freely re-lease or undertake major changes. Data centers can require significant capital to stay current and may need active re-leasing as technology shifts, and a DST is designed to be passive. A long hyperscale lease that needs little intervention during the hold fits the structure better than a facility expected to require ongoing reinvestment.
Diversification, which softens single-asset risk in other sectors, is hard to achieve here given how few offerings exist. An investor may be looking at a single facility leased to a single tenant, which concentrates both the credit and the obsolescence risk in one place. That is a meaningful consideration in a sector where the asset itself can age.
What an accredited investor should scrutinize
Because there is no Baker benchmark or track record to fall back on here, diligence on the specific deal carries even more weight than usual. A handful of questions do most of the work. Start with power: is the facility's electrical supply secured and ample, or does it depend on grid capacity that is not yet locked in? In this sector that question can matter more than the lease.
Then work through the rest. Who is the tenant, what is its credit, and how long is the lease, with attention to whether it is a single hyperscale tenant or a mix of colocation occupants. What does the capital plan look like over the hold, and who pays for upgrades as density and cooling needs rise. How exposed is the building to technology obsolescence, and how re-leasable is it if the current tenant leaves. And, given the rarity of the structure, why was this asset offered as a DST at all, who is the sponsor, and what is their experience with this particular property type.
Who it suits, and who should look past it
Data centers, in the rare cases the sector reaches individual investors, suit a sophisticated accredited investor who understands the technology and infrastructure drivers, can evaluate power and obsolescence risk, and is comfortable with a single specialized asset and no track record to lean on. An investor drawn to the AI and cloud demand story, and willing to do real diligence on the power and capital picture rather than buy the narrative, is the natural fit, if a suitable offering even exists.
It is a poor fit for an investor who wants a proven, widely available 1031 product, dependable and simple income, or the comfort of a long completed track record, none of which this sector offers individuals today. The capital intensity, technology risk, and scarcity of DST options make it a demanding, specialized choice. For most exchangers seeking passive, credit-backed income, a mature net-lease or government-lease asset will be the better-understood path. Data centers are a frontier here, and frontiers reward caution.
Working with Baker 1031
We will be straightforward: data-center DSTs are uncommon, and this is a new sector for our firm. We are not going to quote you a Baker yield or a track record we do not have. What we can do is provide sponsor-agnostic diligence if and when a data-center offering reaches the accredited-investor market, pressing hard on the power, the tenant, the capex plan, the obsolescence risk, and the reason the asset was structured as a DST at all. We are paid to be skeptical on your behalf, and this is a sector that earns skepticism.
Most data-center real estate today sits with REITs and institutional funds rather than in 1031-eligible DSTs, so for many investors the realistic exposure is indirect. The 45-day identification window moves fast, so the time to understand your options, and their limits, is before you sell. We are happy to walk through what is genuinely available to accredited investors, and to be candid when a more proven property type fits your goals better than a frontier one.
View Available Data Center DSTs →
Sources & References
- IRS. Like-Kind Exchanges — Real Estate Tax Tips
- U.S. SEC — Investor.gov. Investor Bulletin: Non-Traded REITs
