🤖 AI Cost Stack & Policy Risk

• AI inference cost pressure is pushing companies toward cheaper open-source models like Zhipu GLM-5.2 and Kimi2.7. If token usage keeps compounding while spend must stay flat, the default model mix becomes the lever. The second-order risk: the U.S. government may soon restrict the use of large models produced in specific regions. 1

🧠 AI Infrastructure Bottlenecks

• DRAM is being framed as the key chokepoint in AI infrastructure. The bottleneck is not lasers, capacitors, power semis, NAND, or HDDs. The logic is simple: if Elon is focusing the Terafab on memory, memory is where the capacity crunch sits. 2

• AI has turned memory into the “bottleneck of bottlenecks.” $MU, Samsung, and SK Hynix allegedly moved from losing money to generating a combined ~$775B in annual profit within five years because data-center demand is bidding up HBM, tightening DRAM supply, and structurally repricing memory. 3

• Big Tech dip-buying may still be painful short term. Nvidia, Google, and Microsoft may already be near historically low valuations, but market liquidity is chasing AI infrastructure choke points like storage and optical modules. Momentum matters in the short run: winners often keep winning. 4

📈 Portfolio Strategy & Risk Management

• Long-term S&P 500 DCA remains a basic retirement floor. Getting rich slowly still counts. Having money at 60 beats reaching 60 with no cushion, so starting early matters more than trying to time the perfect entry. 5

• Precise entries and exits do not have to come from chart-reading. A fundamentals-based playbook can work: buy at a reasonable price, add if it keeps falling, sell decisively when it rebounds to the target price, and avoid falling in love with the position. 6

• Optimism is an edge in U.S. equities. In a long-term uptrend, constant crash fear, “this time is different” thinking, and conspiracy narratives make it harder to hold quality assets, add on dips, and avoid panic-selling near the bottom. 7

• Benchmarking matters. The S&P 500 is up 7.4% YTD, SPMO is up 30%, QQQ is up 15%, and defensive HDV is up 16%. If an account cannot beat the S&P, Nasdaq, or momentum ETFs, automated ETF allocation may be the cleaner trade. 8

• Full margin on top of full allocation is the real problem. If a Tesla position blows up under leverage, the issue is not just Tesla; it is position sizing and risk control. 9

🇨🇦 Canada ETF Allocation & Tax Frictions

• Canadian investors can use VFV as the CAD version of the S&P 500 and HXQ as a CAD-friendly QQQ alternative. VCE and covered-call ETF HDIV are also worth considering after strong performance that can beat the S&P benchmark. In a TFSA, U.S. stock dividends face withholding tax, so structure matters. 10

🛢️ Macro Setup: Oil, Rates & Fed Pricing

• Monday’s open hinges on whether oil converts geopolitical risk into inflation risk. A tanker strike near Hormuz, renewed U.S.-Iran tension, and upcoming jobs data create a clean cross-asset setup: oil up plus $TLT down means pressure on rate-cut expectations and risk assets. 11