- NVIDIA fell 1.51% to $217.44 (≈C$302.00) on September 2, 2026, as a broad semiconductor selloff pulled SMH down 2.05% in a single session.
- Hyperscaler capex from Microsoft, Alphabet, Amazon, and Meta — totalling well over $150 billion USD combined — keeps the AI compute supercycle structurally intact despite daily volatility.
- Toronto’s Cohere, valued above USD $5 billion after its Series D, and the Vector Institute represent Canada’s growing but largely private AI infrastructure, with CSU as the top TSX proxy.
- NVIDIA’s ~35x forward earnings multiple leaves the stock vulnerable to sharp corrections if hyperscaler capex guidance moderates or cloud revenue growth disappoints in late 2026.
NVIDIA (NVDA) fell $3.34 to $217.44 (–1.51%) on September 2, 2026, dragging the VanEck Semiconductor ETF (SMH) down 2.05% to $545.22 as a broad risk-off session hit the AI trade hard. AMD dropped an even steeper 2.36% to $459.61, and Microsoft shed 1.24% to $501.02. In Canadian dollar terms, a single NVDA share now costs approximately $302.00 at the prevailing USD/CAD rate of 1.3888 — a figure that underscores just how much retail investors north of the border are paying for AI exposure.
The Supercycle Thesis: Still Intact, Under Pressure
Today’s pullback is a valuation stress test, not a structural reversal. The AI compute supercycle is anchored in hard capital commitments: Microsoft has pledged over $80 billion USD in data-centre spending for fiscal 2026, Alphabet disclosed $75 billion in capex guidance for the same period, and Amazon’s AWS infrastructure budget continues to expand at a double-digit clip. Meta’s AI infrastructure arm has accelerated GPU cluster deployments across North America and Europe. These are not speculative promises — they are purchase orders, and NVIDIA sits at the centre of almost every one of them.
AI adoption metrics continue to support the thesis. Enterprise generative AI deployment rates crossed 40% of Fortune 500 companies in Q2 2026, up from just 11% in Q2 2024, according to industry surveys. Token throughput on major inference platforms doubled year-over-year, creating sustained demand for both training and inference silicon. Today’s price action reflects profit-taking and macro jitters, not a deterioration in the fundamental demand picture.
Canadian Angle: Cohere, Vector Institute, and TSX Picks
Canada’s AI ecosystem is maturing rapidly, even if the TSX doesn’t yet offer a pure-play NVIDIA equivalent. Toronto-based Cohere — whose enterprise large language model platform competes directly with OpenAI’s API business — completed a USD $500 million Series D in 2025 at a valuation north of $5 billion, with significant backing from Canadian pension capital including OMERS Ventures. The company’s Command R+ model has become a preferred choice for regulated-industry clients in financial services and healthcare. Cohere remains private, but its growth trajectory is a proof point that Canadian-founded AI companies can compete globally.
The Vector Institute in Toronto continues to funnel world-class machine learning researchers into the domestic startup ecosystem — many of them Element AI alumni whose expertise dispersed across the industry after ServiceNow’s 2021 acquisition. On the TSX, Constellation Software (CSU) dropped 2.83% to $3,050.00 today but remains a credible proxy for AI-adjacent software compounding, given its aggressive vertical software acquisition strategy increasingly targets AI-enhanced SaaS platforms. Shopify (SHOP) fell the hardest in the Canadian tech cohort, off 5.12% to $139.82, as investors rotated out of growth names.
Risk Flag: Valuation Compression Is the Real Bear Case
The risk worth flagging today is not demand — it is duration. NVIDIA trades at roughly 35x forward earnings even after this selloff, a premium that leaves little room for disappointment in its next guidance cycle. If hyperscaler capex begins to moderate in late 2026 — a scenario that gains credibility if cloud revenue growth disappoints — consensus estimates for NVDA’s fiscal 2027 earnings could face meaningful cuts. SMH’s 2.05% single-session drop is a reminder that crowded AI trades unwind quickly.
| Security | Price (USD/CAD) | 1-Day Change |
|---|---|---|
| NVIDIA (NVDA) | $217.44 / ~$302.00 | –1.51% |
| AMD | $459.61 / ~$638.17 | –2.36% |
| Microsoft (MSFT) | $501.02 / ~$695.50 | –1.24% |
| SMH (Semiconductor ETF) | $545.22 / ~$756.86 | –2.05% |
| Shopify (SHOP) | $139.82 / ~$194.12 | –5.12% |
| Constellation Software (CSU) | ~$3,050.00 CAD | –2.83% |
For Canadian investors, today’s dip is a moment for discipline, not panic. The hyperscaler capex cycle has years of runway, Cohere and Vector Institute alumni are building globally competitive AI businesses at home, and TSX-listed compounders like CSU offer a less volatile way to participate in AI-driven software economics. But position sizing matters: the AI trade is crowded, valuations are stretched, and the next NVIDIA guidance miss will be felt in portfolios on both sides of the border.
