A data center is an unusually literal answer to a fashionable question: where does the internet live? After the dot-com crash, some of those addresses became distressed property. GI Partners began buying. Between 2002 and 2004, it assembled 25 data center properties with $280 million of equity, according to contemporary reporting. The websites might disappear. The buildings, wiring and possibilities remained.
- GI invests other institutions’ capital across private equity, real estate and data infrastructure.
- Its early property purchases helped create Digital Realty.
- The recurring question: which assets remain useful when technology changes?
The wreckage had an address
Rick Magnuson founded GI in 2001. He had worked in principal finance at Nomura, but his history also included Digital Research and a software services business he founded and sold. His career crossed the boundary that would become central to GI: technology as an operating business, and technology as something that needs financing, equipment and somewhere to stand.
An SEC filing supplies a useful origin detail. CalPERS selected GI after a six-month competitive process and committed $500 million; CB Richard Ellis Investors and management committed another $26.3 million. The mandate connected technology-related real estate with technology operating businesses. This was institutional capital looking for a particular kind of opportunity, rather than a startup pitching an app.
The timing matters. Overbuilding had damaged the data center market before GI made its purchases. What failed first was the economics of the boom. GI’s wager was that selected infrastructure still had a future. Twenty-one early properties became the nucleus of Digital Realty, which went public in 2004. By September 2009, GI reported more than $1 billion in proceeds from the original 25-property portfolio, a 3.7-times gross multiple. That is a reported investment result before fund-level fees and expenses, not a promise about another cycle.
Three ways into the same economy
Today, GI manages three strategies. Private equity backs North American middle-market companies, principally in healthcare, services and software. Real estate buys specialized properties and builds operating platforms. Data infrastructure invests in businesses providing data centers, transport networks, wireless access and connected infrastructure. A building and its operator can share an address while presenting different investment risks.
The published criteria make the distinctions concrete. Private equity allows initial equity investments up to $1.5 billion and enterprise values at entry up to $2 billion. Data infrastructure lists $100 million to $400 million in equity, with meaningful co-investment by limited partners. These are selection parameters, not prices for a customer subscription.
The paying constituency is the institutional investor: pensions, sovereign wealth funds, foundations, investment managers and family offices. Fund VI closed at $3.9 billion in 2021 with investors from 15 countries. The business is managing capital through acquisitions, operating work and eventual realizations. In the broader industry, that work earns management fees and performance participation. A portfolio company’s customers occupy a different layer of the arrangement.
GI’s first dedicated data infrastructure fund closed at $1.8 billion in 2020, above a $1.25 billion target. Its organizing idea favored durable assets and recurring revenue. Specialist firms such as DigitalBridge pursue related territory; large alternative managers compete across real assets. GI’s distinguishing feature is its long practice working across property ownership and operating-company investment.
A tank sensor belongs here, too
Consider Otodata. In September 2026, GI bought a majority interest in the Montreal company, whose systems monitored more than 3.5 million tanks globally. Customers use remote readings to plan work and avoid running out of fuel or other tank contents. Renovo Capital and management retained minority ownership. It is a refreshingly concrete version of the digital economy: someone needs to know what is left in a tank.
The software side has its own practical problems. Egnyte supplies content collaboration, governance and security tools. Its February 2025 investment announcement with GI and TA described more than 22,000 enterprise customers. In healthcare, GI formed Rose BioSolutions in May 2026 from Charles River’s CDMO and Cell Solutions businesses, creating a standalone company serving cell and gene therapy development and manufacturing.
GI’s operating role has a longer example in SoftLayer. It bought EV1 and The Planet in 2006, acquired SoftLayer in 2010 and merged SoftLayer with The Planet. When IBM bought the business in 2013, GI described network upgrades, international expansion and a broader customer base. Acquisitions were part of a sequence of changes to an operating company.
Buying time is part of the job
Private funds have clocks. Companies have projects. Those timetables can diverge. In August 2025, GI announced a Flexential continuation vehicle anchored by Hamilton Lane-managed funds, alongside investment from GI Data Infrastructure. Existing limited partners could receive liquidity while the data center business obtained new capital. Including Morgan Stanley Infrastructure Partners’ earlier investment, Flexential reported approximately $1 billion in primary equity.
Clinical ink offers another example. In October 2026, existing investors, including GI Fund V, exited their prior holdings. Ardan Equity and GI Fund VI invested in a new co-control partnership. The clinical trial technology company continued; the investors and fund vintage changed. Selling an investment and remaining involved can happen in the same transaction.
Copy the question, not the trade
The portable lesson is analytical: separate the customer’s enduring need from the financing and valuation attached to it. Ask who pays, why they keep paying, what must be replaced and how long replacement takes. GI’s stated operating principles emphasize debate and openness to alternative views. Those habits matter when an attractive sector becomes an expensive one.
“It’s trying to distill what you hear into action.”Mike Armstrong, GI Partners / Cool Vector interview, 2026
Durability has conditions. Expensive debt, concentrated tenants, scarce power or delayed construction can spoil a sound demand thesis. In a 2025 interview, infrastructure head Mark Prybutok urged conservative assumptions for developments looking seven years ahead. A long-lived building cannot settle the question of which future workloads will pay its rent. GI’s history makes that question worth asking before the next boom supplies an easy answer.