Oxford buys office again, but only where it can see the leasing

Oxford Properties started selling office in 2018, two years before the pandemic handed the rest of the market its reason to, and the Toronto-based investor has since recycled $14 billion globally out of the sector. Last month the flow reversed. Oxford paid $435 million for a Boston office tower, a deal PRED's records date to Aug. 25 and the firm's first core U.S. office acquisition since 2017. Ankit Bhatt, its U.S. head of investments, told Bisnow the purchase is the start of a buying program rather than a one-off.

What kind of buyer Oxford is being matters more than the price it paid. Bhatt's case to Bisnow rests on vantage point more than on a market call: conditions are changing, but not every investor can declare that this is the moment to go in, because most wait for first movers to validate a market before committing. Oxford's existing office portfolio lets it watch leasing fundamentals and operating activity almost daily, weekly or monthly, and point to momentum as it builds. Owning the buildings is the research.

The geography narrows the strategy. Bhatt named three technology hubs — Boston, New York and San Francisco — and said the firm is evaluating assets with artificial intelligence in mind, weighing how the technology could disrupt the tenants inside each building. His framing of the underwriting question is which tenancies will thrive as AI spreads and which could contract or expand, so that the purchase prices not only today's rent roll but the growth expected later.

That screen is the most consequential piece of the strategy, and the piece most likely to spread. A landlord with a large office portfolio can watch its own tenants' renewals, headcount and space use in a way a buyer working from a broker's rent roll cannot, which turns an AI-exposure test on each tenancy into a defensible way to price the next several years. Our expectation is that tenant-level AI exposure becomes a standard line in office underwriting before this cycle resolves, and that the screen's findings land hardest on commodity buildings whose rent rolls answer it badly.

The 2017 baseline shows how long the firm stayed away. That year's final U.S. office purchase was a pair of Washington, D.C., buildings — 900 16th St. NW and 1101 New York Ave. — bought in a joint venture with Norges Bank Real Estate Management. Bhatt said the decision to start selling came in 2018, two years before the pandemic disrupted the sector, because low interest rates let Oxford maximize proceeds, and the $14 billion it recycled globally went into logistics, multifamily and open-air retail. Logistics grew from nearly nothing in 2017 to roughly $7 billion within seven years.

Set the re-entry against that arithmetic and it looks small. A $435 million tower barely moves a portfolio with more than $86 billion in global assets under management, and it is modest against the $7 billion logistics platform built in seven years. Oxford's own account of the 2018 decision — sell while low rates maximized proceeds — describes a firm that treats office as a duration call it can enter and exit, which is the right lens for judging what Boston means.

It means an information trade. Oxford is buying in markets where it can measure leasing weekly and underwriting tenancies it already knows, and the record of the last decade says it will leave the way it left before if the rate and leasing math turns. The three markets it named are the ones where a landlord's own data is worth the most, which is why the re-entry looks less like conviction in office than conviction in Oxford's ability to price it.

Our own reporting describes the market Oxford is re-entering more precisely than any broker deck. Law firms signed 12.2 million square feet in the first half of 2026, demand that landed in trophy markets rather than across the broad stack, per our Aug. 28 piece on the legal sector's record leasing. Three days earlier we published the Seaport trade that reset the benchmark for fully leased towers while leaving the commodity clearing price unresolved: $879 a foot, the largest pure-play office sale in Boston in five years.

Oxford's shortlist adds a second demand pool, technology, alongside the legal tenancies our reporting tracked, and both sit in the same tier of buildings. That supports the position this publication has taken: office clearing prices are being set by sublease stacks and completed repositions rather than by appraisals, with the spread widening for commodity assets and narrowing for trophy. A first mover paying $435 million in Boston is a data point on the trophy side, and Boston, New York and San Francisco are the geography over which that spread now runs.

The number to watch is not the next purchase price but the cadence. A second and third deal in New York or San Francisco at similar core pricing would make the re-entry a program and would say the trophy spread is closing in more than one market. If Boston stands alone for another year, Oxford will have bought deep knowledge of a market it already knew, which is a respectable trade and a much smaller story than a sector turning.