This 150% Run Is Just the Beginning

Intel has rallied about 150% since March 19 on rising AI infrastructure demand, while AMD rose nearly 19% after strong earnings and guided server CPU revenue growth of more than 70% year over year in Q2. The article argues the AI trade is broadening beyond GPUs into CPUs, networking, memory, and power infrastructure, with hyperscalers planning about $700 billion of AI spending in 2026. Separately, oil has pulled back as Brent fell from about $108 to $101 and WTI from $101 to $96 on hopes for a U.S.-Iran deal, while Canadian oil sands names are highlighted as a separate squeeze trade.

Analysis

The market is starting to re-rate AI infrastructure from a single-node compute story into a systems bottleneck story, and that changes the winners. If agentic workloads scale the way management teams are implying, the constraint shifts from GPU throughput to orchestration, memory access, interconnect, and power delivery — which structurally broadens the trade beyond the obvious GPU leader. That is bullish for CPU vendors and the picks-and-shovels ecosystem, but it also means the next leg of upside is likely to come from names with pricing power and supply tightness rather than pure narrative exposure.

The second-order dynamic is that CPU demand can be more durable than GPU demand because it is embedded in fleet-wide efficiency upgrades, not just new AI deployments. That makes INTC and AMD potentially more leverageable to a long-duration capex cycle than a headline sentiment trade, but it also raises the risk that consensus underestimates how quickly margins can normalize once capacity catches up. In semis, the market usually overpays for cyclical scarcity in the first 2-3 quarters and then compresses multiples before the revenue peak is visible.

On the energy side, the interesting setup is not the headline direction of crude but the widening spread between globally priced benchmarks and stranded/transport-constrained barrels. Canadian producers with low lifting costs and improving egress are effectively getting an embedded margin expansion even if outright oil softens, which makes the trade less about beta and more about basis. That’s why CNQ looks structurally cleaner than higher-cost peers: if the market is pricing a geopolitical fade, the real upside is in names whose realized pricing is still being rerated by logistics.

The contrarian risk across both themes is timing. AI infrastructure spending can remain strong for years, but the stocks can still stall if investors rotate away from the most crowded beneficiaries or if hyperscaler capex converts into slower-than-expected revenue. For oil, peace headlines can hit near-term sentiment quickly, but a sharper reversal in logistics or China demand would re-tighten bitumen economics faster than most investors expect.

Key Decisions for Investors

• Long AMD vs short a basket of mega-cap AI software proxies for 3-6 months: express the view that infrastructure beneficiaries outperform application-layer names as the market re-prices the AI bottleneck toward compute coordination and CPU content.
• Buy INTC on pullbacks only, with a 6-12 month horizon: the trade is best treated as a supply-cycle continuation, not a chase; upside remains attractive if server CPU tightness persists, but use stops if delivery-time data normalizes faster than expected.
• Initiate a long CNQ call spread or stock-plus-call structure into any oil weakness over the next 2-8 weeks: this captures basis-driven upside with limited downside if Brent continues to fade on de-escalation headlines.
• Overweight MRVL, AMAT, KLAC, and LRCX as a basket for 6-12 months: the best risk/reward is in the second derivative of AI capex, where equipment and custom-chip content should benefit even if the market rotates away from top-line GPU enthusiasm.
• Avoid chasing SU/CVE/IMO outright until crude stabilizes; if you want exposure, prefer CNQ or a pair trade long CNQ / short a higher-cost Canadian producer to isolate the margin and logistics advantage.