- NVIDIA fell just 0.44% to $155.46 on November 18, outperforming Microsoft (-2.27%) and Shopify (-4.00%) in a broad tech selloff.
- Hyperscaler capex from Microsoft, Alphabet, Amazon, and Meta has collectively surpassed US$280 billion for fiscal 2026, sustaining GPU demand.
- Toronto-based Cohere and Vector Institute alumni represent Canada’s deepest private AI infrastructure exposure, with no direct TSX pure-play equivalent yet.
- NVIDIA’s ~35× forward earnings multiple is only defensible if data centre revenue compounds above 30% annually through 2028 — a key valuation risk to monitor.
NVIDIA (NVDA) slipped just 0.44% to $155.46 on November 18, 2026 — a notably shallow dip on a day when Microsoft shed 2.27% to $506.24, Shopify dropped 4.00% to $141.90, and the VanEck Semiconductor ETF (SMH) fell 0.39% to $460.10. The price action tells a story: when the broader tech complex wobbles, NVIDIA’s relative strength signals that institutional investors continue to treat it as a non-negotiable infrastructure holding rather than a discretionary growth bet.
The Hyperscaler Capex Engine Is Still Running Hot
The core AI investment thesis rests on a simple but powerful dynamic: Microsoft, Alphabet, Amazon, and Meta are collectively spending at a pace that has no modern precedent in enterprise technology history. Combined capex guidance from these four hyperscalers for fiscal 2026 has surpassed US$280 billion — a figure that flows almost directly into NVIDIA GPU clusters, custom silicon, cooling infrastructure, and fibre interconnects. Each earnings season that reaffirms or raises this guidance is a fresh data point validating the supercycle narrative. November 18’s mild NVDA pullback, against a backdrop of broader selling, suggests the market is not second-guessing that thesis today.
AMD, meanwhile, dipped only 0.17% to $503.41, holding its own as a secondary beneficiary of AI accelerator demand. AMD’s MI300X series continues to win data centre deployments where hyperscalers seek to diversify away from single-vendor GPU dependency — a structural tailwind that has little to do with any single quarter’s earnings beat or miss.
Canadian Angle: Cohere, Vector Institute, and the Quiet TSX Play
Canada’s AI ecosystem is maturing well beyond its academic roots at the Vector Institute in Toronto and Mila in Montréal. Cohere, the Toronto-based enterprise LLM company co-founded by Vector alumni, remains one of the most closely watched private AI infrastructure plays in the country, with its Command and Embed model families gaining traction inside Fortune 500 deployments. Element AI alumni, now distributed across Scale AI, ServiceNow Canada, and a clutch of Series B startups, are quietly building the middleware layer that sits between raw GPU compute and business applications.
On the public markets, Constellation Software (CSU) fell 3.48% to $3,316.93 (approximately CA$4,737 at the current USD/CAD rate of 1.4291) — a reminder that even the TSX’s most celebrated compounder is not immune to tech-sector rotation. For TSX-focused investors seeking pure-play AI exposure, the options remain thin; the most credible route continues to be U.S.-listed names held within registered accounts, or sector ETFs like SMH converted at today’s 1.4291 rate, implying a CAD cost of roughly CA$657.56 per unit.
The Risk No Bull Wants to Talk About
Valuation remains the stubborn counterargument to every bullish capex chart. NVIDIA trades at approximately 35× forward earnings — a premium that is defensible only if data centre revenue growth sustains a compound annual rate above 30% through 2028. Any demand signal that suggests hyperscalers are pausing, digesting, or reallocating capex toward custom ASICs (Google’s TPUs, Amazon’s Trainium, Microsoft’s Maia) could compress that multiple swiftly. The 2.27% drop in Microsoft shares today is worth watching: if it reflects early anxiety about ROI timelines on AI product integration rather than pure macro noise, it is a yellow flag for the entire ecosystem.
| Ticker | Price (USD) | Price (CAD) | Day Change |
|---|---|---|---|
| NVDA | $155.46 | CA$222.10 | -0.44% |
| AMD | $503.41 | CA$719.22 | -0.17% |
| MSFT | $506.24 | CA$723.27 | -2.27% |
| SMH | $460.10 | CA$657.56 | -0.39% |
| CSU (TSX) | — | CA$3,316.93 | -3.48% |
For Canadian retail investors, the November 18 session is a useful stress test: AI infrastructure names held their ground while software and platform stocks bore the brunt of selling. That divergence — compute resilient, application layer soft — is consistent with a market that believes the picks-and-shovels layer of the AI supercycle still has the cleaner risk/reward profile heading into 2027.