The Mushroom Butter at AIR

The first thing on the table at AIR is a small dome of bread and a dish of butter that is the wrong colour.

The bread is fermented cassava, ground and proofed and baked into a loaf with the slightly mochi-ish crumb that root-vegetable doughs produce when they have actually been worked. The butter is darker than dairy. It is cultured at the lab on site with a mushroom XO base, the fermented mushrooms broken down through a controlled bacterial culture until the fat reads more like a small fermented condiment than a churned dairy product. There is no narration. The bread and the butter are placed in front of you and the rest of the meal goes on around them.

That two-item arrival is, after two years and one daypart pivot, where the kitchen makes its real editorial argument. Not in the campus map, the lab tour, or the chef's bio. In a piece of warm bread and a fat that looks like a miso.

I went on a Wednesday with low expectations, mostly because the press cycle around AIR has been louder than the room would, on its own merits, have asked for. Circular gastronomy is the easiest aesthetic position to take and the hardest to actually mean. The phrase belongs to marketing decks and award narratives. The terms that orbit it, fermentation lab and on-site garden and scale-to-stem and cooking school and in-house composting, are the same terms almost every new chef-led concept from a luxury hospitality group has claimed since 2022. Singapore has a long line of restaurants that arrived with this vocabulary and then quietly outsourced the things the vocabulary requires.

AIR has not, on the evidence of the meal, done that.

The room opened on the thirty-first of January 2024 at 25B Dempsey Road. Forty thousand square feet, on a campus that runs the restaurant, a fermentation lab, a cooking school, and a working garden. The chef-partner is Matthew Orlando, Copenhagen via the early Noma kitchen, then Per Se in New York, then Amass in Copenhagen for nearly a decade as the room that did the most to expand the original New Nordic register beyond its first set of moves. His co-founders are Will Goldfarb, the pastry chef whose Room4Dessert in Bali has been one of Asia's quieter arguments for fermentation continued into the sweet course, and Ronald Akili, principal of the Potato Head Family group that operates the campus.

The marketing structure is the marketing structure. The harder thing to evaluate is whether the kitchen has, after two years, actually settled into doing the working version of its own thesis. On the meal I ate, it has.

The bread and the butter

I will not pretend the first bite was complicated. The bread had the soft tang at the back of the palate that fermentation produces in a worked dough rather than a quick bake. The crust gave the right resistance, the crumb pulled apart cleanly, the temperature was correct. The butter spread across the warm surface and immediately revealed what the cultured fat is doing that ordinary butter cannot. The umami arrived earlier. The salt was more rounded. The finish ran longer. The mushroom culture had given the fat a depth the dairy lacked.

I am writing about a piece of bread.

That is, I think, the whole campus shrunk to one bite. If you cannot taste a sustainability programme, it stays on the marketing deck; here you could taste it on the bread. The bread and the butter were the kitchen. By the second bite I had stopped paying attention to what they were and started paying attention to what they were doing, which is the moment a fermentation kitchen has succeeded at the harder work.

The garum on the chicken later in the meal carried the same point at a louder register. The kampong bird was grilled correctly, finished at the table with a sauce the kitchen makes from chicken trim, the parts of the animal a conventional kitchen would have composted, rendered slowly into a savoury liquid that does the seasoning work a jus would. The garum was darker than soy and lighter than fish sauce. It tasted of a chicken the diner could imagine, not of a chemistry the diner had to take on faith.

Two dishes were enough to convince me the lab is real.

What does not work, and why it does not matter much

The Hokkaido potato course was the meal's weakest plate. Twice-cooked, pre-fermented, finished with a fermented-chilli reduction, technically correct, the kitchen's working method on display. But the dish wanted me to know it. The technique read louder than the eating. I noticed the method before I tasted the potato, which is the inversion of what a confident kitchen wants to produce.

That kind of demonstration plate is the structural risk of a fermentation-led menu. The lab is where the kitchen does most of its work, and there is a constant temptation to stage a course that proves the lab is working rather than to cook a course that happens to draw on it. AIR is mostly avoiding the temptation. The bread, the chicken, the rice and the desserts are all dishes that use the lab without performing it. The Hokkaido potato is the course that lapsed.

A diner who is allergic to performance plates should know this in advance. A diner who can absorb one course of self-referential cooking will not find AIR a difficult room.

The next friction is Orlando. The chef is not in the kitchen most evenings. The day-to-day is run by the recently elevated local brigade, a team Orlando promoted from within rather than imported, which is the operational sign that the kitchen is being trusted to interpret the brief rather than execute a single chef's hands. On the evening I ate, the team executed cleanly. The bread was the bread. The garum was the garum. The room moved at the pace the long meal required. But the diner should know that this is the working condition of AIR now: the campus is Orlando's, the kitchen is increasingly the local team's, and the bet is that the brief has crossed the handover.

The third is Dempsey itself. The hill is logistically tiresome at night. The compound's other restaurants, Burnt Ends two doors down, Candlenut nearby, Torno Subito on the same stretch, compete for the same evening. AIR has to earn the trip every time, which is structural overhead the kitchen cannot fix.

The 2025 daypart pivot, read correctly

The most-discussed move at AIR in the last twelve months has been the launch of Breakfast at AIR, a morning service built in collaboration with Le Matin Patisserie and Maxi Coffee. The press read the launch as a softening: a fine-dining room conceding that the dinner-only format would not sustain the campus.

That reading is half right. The campus does need more daypart density to make the square footage work. But the breakfast is not a retreat from the kitchen's editorial position. It is the campus's argument that the operation is large enough to run more than one register. The pastries are Le Matin's pastries. The espresso is Maxi's. The room is open in the morning at a casual register, with the cooking school running parallel sessions, and the dinner continues unchanged in the evening. The campus has gotten larger, not smaller.

The pivot tells the diner what AIR is for now. It is not a tasting-menu restaurant that happens to have a lab. It is a campus that runs a tasting menu in the evening, a casual breakfast in the morning, and a cooking school across the day, with the fermentation lab feeding all three. The economics of a campus that big require the wider format. The kitchen has accepted that without diluting the dinner.

That acceptance is what reads, on the floor, as the kitchen having stopped needing to prove the project. The first year of any campus this ambitious is performative: the lab is shown to the press, the chef gives the interviews, the garden tour is the menu's preamble. The second year is the test of whether the campus continues without the performance. AIR has, on the meal I ate, made the second-year transition. The dishes are not performing the lab. The brigade is not performing the chef. The morning service is not performing the dinner.

What the room is for

AIR is one of the rare campus-scale restaurants in Singapore where the marketing framework has settled into the actual working register of the room. The bread is where you taste the lab, the chicken is where you taste the offcut programme, and the shared rice is the campus's small editorial gesture toward a meal that wants to be eaten together rather than performed. The local team has been trusted to run the brief without the chef's daily hands. The morning service is the campus's quiet acknowledgement that one daypart was never going to be enough.

The mushroom butter, on the warm cassava bread, was where the kitchen made its case at the lowest stakes and the highest credibility: two items, no narration, no theatrics, no chef in front of the table. The lab is real because the bread is real and the butter is real and you can taste the difference between cultured fat and dairy fat without anyone telling you what you are tasting.

That, in a category where most kitchens spend the first three years telling you what they are doing, is the harder version of the project. AIR has, dinner after dinner, stopped telling. The cooking is doing the work the marketing used to do, and that is enough.