Last wk, Muse from $Meta (+12.9%) & co-Pilot overhaul by $MSFT (+4.5%) drove AI infrastructure strength (SOX Index +6.3%) led by processor vendors, $INTC/$AMD/$ARM all +13%. Agentic AI flips GPU:CPU ratio from ~8:1 to ~1:4. SOX+Meta+Microsoft was all the S&P gain. WTI dropped 7.9% last wk while Brent was +0.4% but bond yields surged 10-17 bps across the 3m/30yr curve hitting levels not seen since July-2007 for the 10 year and June-2004 for the 30 year. Yield drivers included: 1) multiple Fed governors messaging higher-for-longer, 2) strong economic data like PMI & capital goods, 3) AI financing concerns following $ORCL (-7% last week) invoking force majeure for their New Mexico datacenter for OpenAI, and 5) weak treasury auction. One of the big use cases for Muse seems to be in cutting monthly bills. This renewed disruption fear for businesses with friction or inertia including telecom $VZ ( -2% last week), insurance $KIE -4%, banking $KBWB -2%, wealth management $SCHW -6%, shopping comparison platforms $AMZN -2% and banned Muse, travel $BKNG -2%, ride-share $UBER -1%, fitness memberships $PLNT -14%, ticket resale $STUB -8% and digital advertising- $GOOGL -2%. These names above have a median return of -1% year-to-date with half of them down ytd. In the disruption fears category above, Google is the one that I am most bullish on. I believe they are very close to launching a new model that will challenge the frontier leaders of Anthropic and OpenAI. In addition I believe all LLM companies will launch their own competitor to Meta Muse including Google. OpenAI DevDay 2026, is taking place on Tuesday. I ultimately believe that consumers will use an AI agent sitting on their smartphone from $AAPL or the Android ecosystem from Google while enterprises will default to the "safety" of co-Pilot on M365 from Microsoft. I believe the AI agent will pick the underlying LLM most suited for the task and commoditize that layer with the value accruing to the infrastructure layer. On a fundamental basis, CQ3 earnings appear solid, especially for AI infrastructure companies. Implied S&P revenue/EPS for Q3 remains strong at 14%/35% y/y with the Atlanta Fed GDPNow forecast at 5%. The question will be how companies react to widely expected beat and raise quarters. Q2 results were not taken well with the Nasdaq -3% and Semi Index -21% during the month of July. The stock reaction to widely expected strong quarterly results/guidance from $MU on Wednesday should give us some clues. From a bigger picture standpoint, the R2K, which is the most sensitive to rates, was down 0.8% last week. While I am impressed by the S&P, the gains were narrow and powered by the AI infrastructure names I am particularly bullish on: microprocessors within semiconductors on the advent of Agentic AI (see initial post on 3/31) and Meta (see initial post on 9/7.) Microsoft which I posted about last Sunday is also a favored name as a “safe” way for corporations to access AI though their 27% OpenAI ownership concerns me. Looking back at history, 20% or greater drawdowns in the S&P have been preceded by or accompanied by the combination of 1) surging bond yields, 2) surging oil prices and 3) Fed tightening cycle. This includes: 1) 1973–1974 OPEC Oil Embargo and Stagflation, 2) 1980–1982 Volcker Rate Shock, 3) 1990 Gulf War, 4) 2007–2009 Global Financial Crisis, and 5) 2022 Post-Pandemic Inflation Crisis. It is possible that the massive AI infrastructure spend & resultant S&P EPS growth of 35% y/y can offset this. It certainly has so far. However, the three middle episodes above had the following positive technological developments as well: 1) the PC revolution started in the late 1970s, 2) the Internet in the mid-1990s and 3) smartphones in the late 2000s. As a result, until bond yields stop their surge, I am remaining very selective on the individual names & sectors I favor. I am therefore watching closely the datacenter pushouts, the developments in Iran and the impact on oil prices. All the best in the week ahead.