#38

1. In 2023, Toronto introduced the Multiplex Bylaw, which allowed any owner of a residential property to redevelop their property as a three-storey multiplex with up to four units, up to 19m in depth and 10m in height. This was later updated to allow those multiplexes in a semi-detached configuration, sharing one party wall. Earlier this year, a Zoning Bylaw Amendment pushed those permissions up to six units and 10.5m in height, though it was only adopted in nine of the city’s 25 wards.
   
   These are starting to get built, though mostly still in the old City of Toronto, where parking-free rents are sufficiently high to make them pencil. Over time, as these permissions continue to improve, I think City Planning should start to review the side yard setback requirements. Currently, these are 0.45m if there are no side windows and 0.9m if there are side windows. They reinforce a detached, suburban streetscape and do too little to improve the beauty and walkability of our neighbourhoods.
   
   The reason side yard setbacks exist, of course, is to minimize or mitigate impacts on neighbouring properties. I think there’s a way to move toward no side yard setbacks for multiplexes while taking those concerns seriously, and without sending every project through a lengthy and uncertain planning process.
   
   City Planning should draft a templated Limiting Distance Agreement, that if entered into by a multiplex developer and his neighbour, allows for the multiplex to be built up to the side lot line. Developers could then negotiate with neighbours, pay them, or whatever, but ultimately get buy in from the only affected party.
   
   To keep things simple, this template should itself be very simple, and should grant an automatic as-of-right elimination of the side yard setback requirement.

2. City Council also recently voted to allow some more retail shops in neighbourhoods. If you live in Toronto, you probably know of one or two old corner stores embedded in residential streets, but almost none that have opened recently. That’s because new corner stores became a prohibited use in Neighbourhoods in 1959, and have basically been frozen in place until this year.
   
   The new policy will allow small retail shops to open on corner lots on select so-called “community streets” within Neighbourhoods, and again—as with the six-unit multiplex permissions—only in a small share of Toronto’s wards, primarily those in the old City of Toronto and East York.
   
   I think these policies are still too restrictive in a bunch of ways, but, in a similar vein to the point above, I’d start by asking City Planning to allow these shops on any corner lot or on any lot adjacent to an existing retail shop. That way, you can imagine a corner store, then a shop next to the corner store, then another next to that, and so on, opening up. You start to get a little run of shops instead of a single, lonely holdout on the block.
   
   This initiative as whole should also work to make our neighbourhoods more beautiful and walkable.

3. I’ve just launched ticket sales for the 2026 Missing Middle Summit. We’re doing it at a really cool new venue and have some great new speakers lined up, including people who are actually delivering multiplex and small apartment projects under these new permissions. If you’re interested in building at this scale, as the rules continue to improve, you’ll want to check it out. It’s happening on May 13th, 2026, and you can buy your tickets at missingmiddlesummit.com.

4. I think we’re less than ten years away from AI being good enough to do all the design work we typically do through entitlements, including site plan control and approval. That means most of the heavy lifting currently done by an architect, landscape architect, structural engineer, civil engineer, MEP engineer, and transportation planner. Realistically, we might even be closer to five years for entitlement-level work, and maybe ten years away from AI being good enough to take this all the way through to working drawings. That’s pretty wild, and it makes you wonder where some of that unlocked value will ultimately accrue. As I argued in my last newsletter, I think good architects, engineers, and planners end up getting paid much more in this world, certainly in the medium term, because the leverage on judgment, taste, and coordination goes up dramatically, and because regulators will almost certainly require a licensed human to remain in the loop and take professional responsibility for the final product.
   
   From the perspective of a developer, this isn’t just about saving money on fees (though it will do that). It is also about enabling much more iteration, like 100x more, because iteration becomes cheap and instant. Today, if we want to explore a different massing, unit mix, parking approach, or servicing concept, we are talking about four to six weeks and many thousands of dollars to push a “simple” change through all of those specializations, with multiple rounds of emails, markups, and coordination calls. For that reason, we mostly keep iteration to a minimum and keep it between us and our architect until we lock in a direction and bring everyone else into the mix. When we are responding to City comments, however, we do need updates from the whole team for submission.
   
   In an AI-first workflow, you can imagine treating that whole stack as something you can re-run on demand: “show me ten options that take advantage of new Major Streets permissions, include minor variances with a high probability of approval, optimize the unit mix for this submarket, minimize structural complexity, and achieve full scoring for MLI Select’s energy efficiency criteria.” You could run that prompt dozens or hundreds of times before you ever file for SPA or lock in a direction, and then re-run it just as quickly to respond to City comments.

5. As more and more of the development process gets automated, I think the important edge will increasingly be cost of capital. If AI compresses the gap in execution quality, design speed, and entitlement strategy, then a lot of what used to separate good developers from the pack becomes more widely available. Everyone will have access to better design tools, faster coordination, and smarter underwriting. At some point, they might even have access to the same robot trades! What they will not all have is the same balance sheet, the same relationships, or the same cost of debt and equity.
   
   If your cost of capital is lower, you can do a few very simple but very powerful things. You can pay a bit more for land and still hit your return thresholds. You can hold land longer through entitlement risk and market cycles without getting squeezed. You can structure deals in ways that are more attractive to partners. In a world where pro formas start to converge because the tools are so good, the spread on your interest rate and your required equity return starts to matter a lot more.
   
   That also suggests where developers should be investing more of their time and reputation over the next decade. A secondary edge might be political access and influence, which feels a bit unsavoury but is nevertheless, I think, true.