The macro renewal-wave story has been told for three years. This week it stopped being macro. The Bank of Canada's 2026 Financial Stability Report put the final wave of pandemic-era fixed mortgages in one clean number: about 12% of all outstanding Canadian mortgages, the last cohort of five-year fixed loans originated at 2020 to 2021 lows, will renew over the next 12 months at an average payment increase of about 15%. And the same report estimates that roughly 4% of 2027 renewers nationally, and about 9% in the Toronto area, may not qualify to refinance at current rates and home values. Those borrowers cannot shop the market for a better rate. Their only route to a new term is signing what their existing lender offers.
The lender-side data confirms the pressure has already moved from projection to portfolio. Bank-held mortgage arrears hit 0.28% in April, a 12-year high, up 26% year over year across 13,752 files. GTA 90-day mortgage arrears are roughly quadruple their post-pandemic lows. The renewal wave is no longer a chart in a Financial Stability Report; it is a rising line in every Canadian lender's monthly delinquency deck. Retention and restructuring decisions have shifted from customer-experience conversations to credit-risk conversations, and both must be handled by the same workflow at the same speed.
The phrase "mortgage prison" is media framing. The operational reality behind it is precise. A borrower who purchased at 2020 to 2021 pricing with a modest down payment and is renewing today into a materially higher rate may face two constraints simultaneously: the new payment consumes a materially larger share of gross income, and the property's current value plus current stress-test requirements no longer support refinancing to a new lender. That combination collapses the borrower's optionality to whatever the existing lender is willing to offer. It also concentrates the lender's decision to just three options: match a competitor rate that the borrower cannot actually access, offer a restructured term that keeps the file performing, or accept that the file is on a path to arrears and manage the workout early rather than late.
The volume math is what makes this a decisioning problem rather than a retention-team problem. Approximately 60% of all outstanding Canadian mortgages will be renewed in 2025 or 2026, with the Bank of Canada estimating that about one-third of all mortgage holders will see payment increases by the end of 2026. Five-year fixed borrowers renewing in 2026, who account for about 40% of outstanding mortgages, face average payment jumps of 15 to 20%. Every one of those files needs a real-time policy view: does the borrower qualify at current rates and values, can they absorb the payment increase, does an amortization extension keep the file performing, is a blended-and-extended rate the right economic answer, or does the file need to move to workout before the arrears clock starts? Handling that decision manually at portfolio scale is impossible; handling it with a generic AI overlay is dangerous.
The wrong retention decision is not just a lost customer. On a renewing borrower who is cash-flow constrained but performing, a rate match that beats a competitor by 15 basis points can preserve a relationship worth thousands over the term. A restructure that extends amortization on the same file can keep it performing when a straight renewal would have pushed it into arrears. A workout referral 90 days before the reset date preserves recovery value in a file that would otherwise degrade for 6 months before the lender sees the problem. Each of those decisions requires reading the current-rate, current-value, current-income, current-payment, and current-consent state simultaneously and applying the institution's own policy to the specific facts of the specific file.
Foundation models give every Canadian lender the same baseline ability to read a renewing file. That reading capability is a commodity. The signal that separates a defensible retention, restructuring, or workout decision from a generic recommendation is the lender's own credit policy and the lender's portfolio managers' historical corrections applied across thousands of renewal decisions. FundMore AI is policy-trained per lender; the specialized signal stays with the lender, not the vendor, and it compounds with every decision the system processes. That is what converts a book that would generate reactive workouts into a book that produces proactive retention and structured restructures at scale.
The integration reality matters too. Renewal decisioning is not a greenfield build; it is a workflow change on top of an existing origination and servicing stack. FundMore deploys as agents over a lender's existing LOS, no rip-and-replace, no 12-month IT project, no asking a bank or credit union to bet its core on a vendor that also originates, funds, or brokers loans. FundMore never originates, funds, or brokers, which means the institution's borrowers stay the institution's borrowers, and the data the institution processes through FundMore is governed by the institution's policy. The renewal-wave response has to be executable in the current fiscal year, not the next; the deployment posture has to match.
Digital twinning matters once renewal decisioning runs at production volume. Lenders need to test rate-matching thresholds, restructure eligibility rules, and workout triage models against realistic portfolio behaviour before deploying to live files. Privacy-safe synthetic twins derived from platform data, disclosed as model-derived rather than presented as real customer records, answer to PIPEDA, Quebec's Law 25, and OSFI's model risk expectations simultaneously. Lenders that can test policy changes on a twin before pushing them to production will move faster on the renewal file; lenders that test on live borrowers or refuse to test at all will spend 2027 explaining unforced errors to their credit committee.
The Bank of Canada's message this cycle has been consistent: the pressure is concentrated, not systemwide, and the peak of the renewal wave passes by mid-2027. That is the macro read. The micro read for every Canadian lender is different. The 12% of outstanding mortgages renewing in the next 12 months at 15% payment jumps live on your book. The 4% to 9% of Toronto borrowers who cannot refinance elsewhere are renewing with you by default. The 12-year high in arrears is showing up in your delinquency deck. The primary-lender relationship at renewal is stickier than assumed in the last cycle, and decision quality on each retained file matters more.
Retention has moved from a servicing metric to a credit-decisioning problem. The institutions that build a renewal decisioning stack that produces three outcomes per file at policy speed, a defensible retention offer, a structured restructure option where supported, and a clean workout referral where required, will exit this cycle with a stronger book than they entered it. The ones that leave retention to the rate-match desk and work out to the collections team will spend the next 24 months paying for the gap.
A: About 12% of all outstanding Canadian mortgages, the last cohort of pandemic-era five-year fixed loans, will renew over the next 12 months at average payment increases of roughly 15%, according to the Bank of Canada's 2026 Financial Stability Report. More broadly, roughly 60% of all outstanding Canadian mortgages renew in the 2025-2026 window, with about one-third of holders seeing payment increases by the end of 2026. Five-year fixed renewers in 2026 are the highest-pressure cohort at 15 to 20% average jumps.
A: Per the Bank of Canada's Financial Stability Report, approximately 4% of 2027 renewers nationally, and about 9% in the Toronto area, may not qualify to refinance at current rates and current home values. Those borrowers cannot shop competitively for a new lender; they are effectively locked to their existing lender's renewal offer. A further 10% decline in home prices would push those numbers to 7% nationally and 12% in Toronto. Operationally, it means the primary-lender relationship at renewal is stickier than the last cycle, and retention decisions carry more strategic weight.
A: Bank-held mortgage arrears hit 0.28% in April 2026, a 12-year high, with 13,752 files in arrears, up 26% year over year. GTA 90-day mortgage arrears are roughly quadruple their post-pandemic lows at approximately 0.24%. Arrears remain low by historical standards, but the direction and pace of change are what matter for reserve planning and workout capacity.
A: Consensus expects a hold at 2.25%, the sixth consecutive pause, with bond markets pricing roughly a 7% probability of a 25-basis-point hike. Most Canadian bank economists expect the policy rate to hold through 2026, with the first move in Q2 2027. For renewal borrowers, that means the operating environment they will renew into for the rest of 2026 is largely set; the payment-shock math does not materially change unless second-round energy-price effects force the BoC's hand.
A: FundMore deploys as agents over a lender's existing LOS, no rip-and-replace, working as an extension of current systems. FundMore AI is policy-trained per lender: each lender's credit guidelines and each lender's underwriter and portfolio-manager corrections shape the retention, restructure, and workout decisions the system produces over time. Foundation-model reading capability, current-rate, current-value, current-income data, and the institution's own policy signal combine into a decisioning layer that produces three defensible outcomes per renewing file at policy speed, rather than a single reactive rate match.
A: Three priorities. First, segment the next-12-month renewal pipeline against current-value and current-income data now, rather than waiting for each file to hit its reset date; borrowers who need a restructure or a workout referral are identifiable 90 to 120 days in advance. Second, put a policy-trained decisioning layer in front of that pipeline so retention, restructuring, and workout decisions run at production speed with a defensible audit trail. Third, align the servicing, credit, and workout teams around a single decision output per file, rather than three sequential handoffs that waste time and information.