Bruce 5 Enters Refurbishment: Ontario's Nuclear Programme Moves Into Its Second Half — and Its Hardest Test

Bruce Power removed Unit 5 from the grid on 1 November 2026, opening the fourth Major Component Replacement (MCR) outage in a six-unit programme that will keep overlapping refurbishments running at the Bruce site in Ontario until 2033. Unit 5 is not the headline project — Units 6 and 3 have already returned, Unit 4 is mid-outage — but its start is the point at which the programme stops being a demonstration and becomes a production line with a fixed end date.

What is actually being replaced

The MCR is a mid-life rebuild of a CANDU reactor. Per Bruce Power, each unit has its steam generators, 480 fuel channels (pressure tubes) and calandria tubes, and 960 feeder tubes removed and replaced, adding 30–35 years of operating life. The work is done with the reactor face opened — the reactor building becomes a construction site for roughly three years, with fuel defuelled, bulkheads installed and heavily shielded tooling deployed to cut out activated components.

That sequence matters commercially. Unlike a welding or mill rebuild, the outage does not scale by adding shifts at the end; the critical path runs through a fixed number of pressure-tube and feeder removals and installations that can only be accelerated by better tooling or parallel working. This is exactly where Bruce's incremental gains have come from — the company cites the first use of robotic tools on a reactor face during the Unit 3 outage as the basis for its schedule improvement.

The performance record Bruce is building on

Sources: Bruce Power releases (2 Nov 2026; 8 June 2026); World Nuclear News; CNSC regulatory hold-point page; TC Energy Q2 2026 filing. Schedule windows are CNSC's published refurbishment ranges, not commitments to return to full power.

Two points in that table deserve more scrutiny than the press releases give them.

First, the "ahead of schedule" claim is measured against a renegotiated baseline. Unit 3's original committed return date was January 2027; it came back in June 2026. But the gap between Unit 6's January 2020 start and its September 2023 return was nearly four years, against a roughly three-year plan for later units. Much of the Unit 3 improvement reflects learning curve and tooling rather than a programme that was ever going to run that long. The relevant benchmark for Unit 5 is Unit 4, not Unit 6.

Second, the two owners describe Unit 4 differently. Bruce Power says the unit is "tracking ahead of the Unit 3 MCR"; TC Energy — which holds an equity interest and reports to the SEC — describes Unit 4 as "on time and on budget." That is not a contradiction, but it is a difference in emphasis worth noting: the partner with disclosure liability is choosing the more conservative formulation.

Why Unit 5 is economically different from Units 3 and 4

The start of Unit 5 is not a repeat of previous gates; it changes the incentive structure around the remaining work.

Under the arrangement with Ontario's IESO, savings against the locked-in cost estimate are shared with ratepayers — 50% for the first unit, 75% for the second through fourth, with incremental sharing for the fifth and sixth. Unit 5 is the fourth, so the 75% tier applies. Any cost underrun Bruce Power delivers on Unit 5 flows disproportionately to electricity consumers, and the company's own framing — "privately backed" MCR, savings returned via the IESO contract — is aimed squarely at that mechanism.

The gates also still exist. The IESO may decline a refurbishment if the estimate exceeds the base-case cost by more than 30%, and economic off-ramps remain before Unit 7 and for Unit 8. Approving Unit 5 effectively closes the fourth of six decision points; the remaining two are the ones where the option value of not proceeding is highest. Starting Unit 5 is therefore as much a governance event as an engineering one.

The supply chain is the real constraint now

The component contracts were placed years ago and are largely fixed:

• Steam generators: BWXT Canada designed and supplied the units (a contract originally valued around CAD 640 million for 32 generators); installation is handled by the Steam Generator Replacement Team (SGRT), a 50/50 Aecon–SGT joint venture, under a contract covering Units 5, 7 and 8 reported at roughly CAD 700 million.
• Fuel channels and feeders: Shoreline Power Group holds a separate long-term contract, reported at around CAD 1.3 billion over ten years, covering Units 4, 5, 7 and 8.
• General contracting and trades: a large, specialised unionised workforce that Bruce Power says sustains 22,000 direct and indirect jobs and about CAD 4 billion in annual economic benefit; the company states 95% of total spending is in Canada.

This is the crux for suppliers and for anyone watching the programme. The binding constraint through 2033 is not components — those are contracted — but the availability of skilled trades capable of working on a reactor face under radiological conditions, and the tooling that lets them work faster. With Units 5, 7 and 8 overlapping through 2033, the same crews and the same specialised contractors rotate between units. Any slippage on Unit 5 does not just delay Unit 5; it compresses the window before Unit 7 mobilisation in 2028.

That makes the current phase a test of industrial capacity rather than of nuclear engineering. It is also the reason the 15% cost reduction claimed on Unit 3 matters more than its headline number: if that learning curve is genuinely transferable, it is the only mechanism by which a five-outage sequence running to 2033 stays affordable.

The Ontario system context

Ontario's grid is unusually exposed to this schedule. The province's phase-out of coal and its reliance on nuclear baseload mean that refurbishment outages remove large blocks of firm, low-carbon generation precisely as demand from electrification and data centres rises. Bruce 5 itself is a substantial single unit, and the overlapping outage pattern means the site's available capacity steps down and back up repeatedly over the next seven years. The IESO's planning assumes those units return on schedule; that assumption is now being tested in real time.

One further layer: Bruce Power is also a major supplier of medical isotopes, particularly cobalt-60 and high-specific-activity lutetium-177, produced through reactor irradiation. Refurbishment outages interrupt that irradiation capacity, which is why Bruce Power frames isotope supply as part of the MCR case rather than a side benefit — a point with direct implications for radiopharmaceutical supply chains that rarely appear in electricity-market commentary.

What to watch

• Unit 4's return date. The CNSC's published window runs to Q4 2027. If Unit 4 returns materially ahead of that — as Unit 3 did — the learning curve story is real. If it returns on time rather than early, the improvement trend has plateaued.
• Hold-point cadence on Unit 5. The CNSC requires approval at four regulatory hold points (fuel load, reactor start, and crossing 1% and 35% full power). The pace at which these clear is the cleanest public signal of construction progress.
• Cost disclosure versus the IESO estimate. Unit 5 sits in the 75% ratepayer-sharing tier. The size of any eventual refund is the most direct measure of whether Bruce Power's process improvements translate into money.
• Units 1 and 2. Ontario is evaluating a second MCR for the two units returned to service in 2012. A decision there would extend the Bruce programme beyond 2033, and would be the strongest possible verdict on the refurbishment model itself.

The analytical judgement

The significance of 1 November 2026 is not that another reactor went offline. It is that a privately financed, ratepayer-incentivised refurbishment programme has now passed its midpoint with a credible record of delivering early and under budget, and is entering the phase where its own success — three overlapping outages, one contractor pool, one end date — creates the conditions for its most likely failure mode. Bruce Power's claim that the remaining work is "a magnitude of work that no other utility in the world has faced" is promotional, but not exaggerated. Whether the industry treats CANDU refurbishment as a repeatable industrial product will be settled by Units 5, 7 and 8, not by the two that already came back early.