TL;DR
- Obsolescence now has dates. England and Wales require EPC B on privately rented commercial buildings over 1,000 sqm from 2031, and the EU's EPBD obliges every Member State to lift its worst 16% of non-residential stock by 2030.
- Conversion is a minority exit. Gensler screened more than 1,300 candidate buildings and found roughly 25% suitable, and the winners are towers with the "wrong" office geometry.
- Ownership decides more than engineering. Citi completes a whole-tower retrofit at 25 Canada Square in 2026; Tour Montparnasse, split across 34 co-owners, watched a EUR 727.7 million programme lapse in July 2026.
Frankfurt's office vacancy reached 11.1% in the second quarter of 2026 and La Défense sat near 14.5% through 2025. Neither market is short of tenants. Both are long on the wrong buildings. A European tower finished between 1960 and 1990 now carries a structure good for another century, services and a facade that are already dead, and a floor plate designed for a working day that no longer exists. The question is no longer whether it is obsolete. It is which of three exits it qualifies for.
That question is worth answering early, because the three exits price differently and only one of them is cheap. Carbon and the tower handles the carbon arithmetic. This piece handles the balance sheet.
By the numbers
- 11.1%Frankfurt office vacancyQ2 2026, up from 10.4% in Q1JLL
- 14.5%La Defense office vacancyabout 536,000 sqm empty in 2025Knight Frank via Qoridor
- 25%Towers suitable for residential conversionof more than 1,300 buildings screenedGensler
- 727.7EUR mTour Montparnasse works budgetApril 2026, excluding tax, up from EUR 624mImmomatin
- 387,000sqmWarsaw office stock withdrawn37 buildings, 2020 to 2025Colliers via NowaWarszawa
- 2031EPC B deadline, England & Walesprivately rented buildings over 1,000 sqmUK government
The compliance calendar replaced the market opinion
Obsolescence used to be a leasing agent's judgement. It is now a date in a statute. On 18 June 2026 the UK government confirmed that privately rented commercial buildings over 1,000 sqm in England and Wales must reach EPC B by 2031, "where cost effective", and dropped the interim EPC C milestone previously proposed for 2027. Smaller buildings stay at EPC E. In the EU, Article 9 of the EPBD recast obliges each Member State to lift its worst-performing 16% of non-residential buildings above a national threshold by 2030, and 26% by 2033.
The exposure was sized before the deadline moved. Savills found in March 2022 that 85% of office stock in the major UK office markets was rated EPC C or below, with 800 million sq ft below B, at an indicative cost upwards of GBP 40 per sq ft and as much as GBP 63 billion across seven UK cities. Savills also put the share of UK office buildings already at B or above at only around 10%. A 1970s tower is on the wrong side of every one of those lines, and the economics of height make each square metre of the fix dearer than in a mid-rise block.
Vacancy is not a market problem, it is an age problem
The headline vacancy numbers hide the split. JLL put Frankfurt at 11.1% in Q2 2026, roughly 1.32 million sqm empty, up from 10.4% a quarter earlier, while prime rent rose to EUR 55 per sqm per month. Scarcity and glut in the same market, at the same time. Prime space is bid up; secondary space is not let at any price.
Citi challenged us to make a 20-year-old tower compatible with its vision for the future of work.
Yasmin Al Ani Spence, Director, WilkinsonEyre
Paris shows the same shape at district scale. Knight Frank counted about 536,000 sqm vacant at La Défense in 2025, close to 14.5% of the stock, against a Paris central business district that has run near 2%. The district operator's own answer is not to relet it. Paris La Défense has set a programmatic framework targeting a 7% to 8% vacancy rate by 2040 and roughly 5,000 homes created out of obsolete offices, plus hotel capacity taken from 3,000 to 5,000 rooms. That is a plan to shrink the office district, written by the office district.
Warsaw has already run the experiment. Colliers data reported in June 2025 counted 387,000 sqm across 37 buildings withdrawn from the Warsaw office market between 2020 and 2025. The split of what happened next is the most useful number in this piece: 37% changed function, mainly to residential, 27% were demolished outright, 26% were modernised, and 10% were sold or kept for owner use. Roughly a quarter of withdrawn stock came back as an office. See Warsaw: how a mid-size capital built a skyline for how that stock arrived in the first place.
Only about a quarter of towers convert, and the good ones look like bad offices
Office-to-residential is the most discussed exit and the least available. Gensler, having scored more than 1,300 candidate buildings across 130-plus cities, reports that only 25% are suitable candidates, and that a conversion typically lands about 30% below the cost of new construction. Its screen inverts the usual quality hierarchy. Shallow floor plates put every room near a window. Small cores leave usable perimeter. A typical Class C office at roughly 12 ft floor-to-floor yields about 11 ft of residential clear height. The features that made a tower a poor office in 2019 make it a viable apartment building in 2026.
Centre Point in London is the proof. Built between 1963 and 1966 to R. Seifert and Partners' design, 34 storeys and 117 m, Grade II listed in 1995, it was acquired by Almacantar in 2011 and reopened in 2018 as 82 apartments with the precast exoskeleton restored and the first two floors given over to amenities. Almacantar's own framing is blunt: as a place to work, the building had become unfit for purpose.
Frankfurt is doing it at volume rather than as a trophy. The 18-storey office tower at Lyoner Strasse 14 in Niederrad is being revitalised rather than demolished into 364 apartments, 220 micro-living units and 144 serviced, with the structure retained and the facade renewed. In the Westend, a 1975 office ensemble is being remade by Groß & Partner as a 27-storey residential tower of around 130 apartments alongside a 20-storey hotel. Warsaw's case is smaller and later: Saski Point at Marszałkowska 111, completed in 2000, bought from CA Immo for about EUR 23.9 million and permitted on 19 January 2026 for roughly 45 to 60 apartments of 50 to 150 sqm.
In 2021 we saw more than 90% of leasing activity being for the best quality office accommodation in some locations.
Mat Oakley, Head of Commercial Research, Savills
The deep-plate, big-core towers of the 1980s and 1990s fail the same screen, and the arithmetic of why is visible in Warsaw's Intraco. Built in 1975 at 107 m across 39 floors, it is being demolished by Polski Holding Nieruchomości and replaced by a tower of the same 107 m carrying 23 floors, for completion in 2030. Sixteen floors of saleable area disappear into modern ceiling heights. PHN's stated reason is that the building "no longer meets today's technical standards and market expectations". Ilmet, PZU Tower and Pekao Tower are on the same list.
Strip back to the frame, and let the cap table decide
Where conversion fails, the surviving route is to keep the concrete and replace everything else. At Canary Wharf, Citi bought 25 Canada Square (completed 2001, 200 m, 1.2 million sq ft) for GBP 1.2 billion in 2019 and chose refurbishment over demolition, a decision it says avoids more than 100,000 tonnes of embodied carbon, equivalent to running almost 22,000 family cars for a year. WilkinsonEyre's scheme reuses elements of the original facade. Work started in July 2023 for completion in 2026. The reported cost travelled from GBP 100 million to over GBP 300 million to around GBP 1 billion, per Fortune in January 2025 citing the Financial Times, a figure Citi disputed. One owner, one occupier, one decision, and the number still moved by an order of magnitude.
Tour Montparnasse (1973, 210 m, 59 floors) has the same engineering problem and none of the governance. The works budget reached EUR 727.7 million excluding tax by April 2026, up from EUR 624 million the previous summer, spread across 34 co-owners led by LFPI, MGEN and Axa. The tower was emptied on 31 March 2026. On 21 July 2026 the co-owners approved only a first phase of asbestos removal and facade stripping to the fourth floor, and the full programme was declared lapsed, with disputes over hotel floors, surface exchanges and rooftop rights unresolved. The 2019 permit runs to 26 November 2026. The engineering was never in question. The cap table was.
The second life can pay, and there is a number for it
The Frankfurt precedent is the one owners should read. Deutsche Bank bought its twin towers at Taunusanlage 12 (built 1979 to 1984, 155 m, about 60,000 sqm) in 2007 for roughly EUR 271 million, then spent an estimated EUR 200 million gutting the climate, water and lighting systems and re-glazing with opening windows. Energy use and CO2 emissions fell by at least 50%, and the towers became the first high-rises certified LEED Platinum for existing buildings by the US Green Building Council. In March 2011 the bank announced a sale to DWS Investments for EUR 600 million. About EUR 471 million in, EUR 600 million out, four years. A strip-back retrofit is not automatically a value trap. It is a development, and it is priced like one, with the same execution risk that engineering the supertall describes.
Winners & losers
Who gains:
- Owners of shallow-plate, small-core towers from the 1960s and 1970s, whose "obsolete" geometry is exactly what the residential screen rewards.
- Single-owner, single-occupier towers, which can approve a EUR 200 million to GBP 1 billion programme in one meeting.
- Contractors and asbestos specialists, since even the lapsed Montparnasse scheme funded a two-year first phase from September 2026.
- Cities running conversion at district scale, where Paris La Défense books 5,000 homes and Frankfurt books 364 apartments per tower.
Who pays:
- Fragmented co-ownerships, where any single blocking interest converts a solvable engineering problem into a stranded asset.
- Holders of deep-plate 1980s and 1990s stock, which fails the conversion screen and faces a full retrofit or demolition.
- Anyone underwriting floor count rather than floor height, as Intraco's 39 floors becoming 23 at the same height demonstrates.
- Landlords of sub-B stock in England and Wales, now five years from a statutory letting restriction.
What lenders and valuers now write down
The repricing arrives through the valuation standard rather than the credit spread. RICS published the second edition of its bank lending valuation professional standard on 9 October 2025, effective 1 January 2026, together with practice information on the prudently conservative valuation criteria introduced under Basel 3.1 and implemented in the EU through the amended CRR III. RICS global standards already require sustainability and ESG to form an integral part of the valuation approach and reasoning, with MEES and EPCs named in the UK supplement. A 1980s tower with no credible retrofit route is no longer merely hard to let. It is harder to value, and therefore harder to lend against.
Three dates decide the next phase. Montparnasse's 2019 permit expires on 26 November 2026, which forces the co-owners to either re-permit or abandon. Citi completes 25 Canada Square during 2026, giving the market its first fully priced whole-tower European retrofit benchmark. And the EPBD's first non-residential threshold binds in 2030, one year ahead of England's 2031 EPC B date, which means European valuation evidence should arrive before UK enforcement does. Owners screening their stock in Q4 2026 still have time to choose an exit. Owners who wait until 2029 will have the exit chosen for them. The full picture of what stands and what is rising is in Europe's tall-building map 2026, and the rest of the argument runs through this Skyscraper Day Dive.