Financing announcements point to more accelerator demand, supporting semiconductors. About $200bn in announced Meta, Anthropic/Broadcom and SpaceX/Nvidia financing likely drives $400bn in XPU purchases. Data centers burn cash until lit, then cash flow 20-40% annually; AI spend rising from 0.1% to 0.5-1% of current payroll could balance the buildout if agents replace roles. Sustainable token demand is the bet, but smaller local models are the bigger risk, though not an immediate concern. x.com  … Show more

Semiconductors could stay undersupplied for a long time as more capable AI per dollar creates more uses and drives compute demand. Micron and NVIDIA trade at low forward P/Es because investors fear peak cyclical earnings, yet current earnings show no slowdown. Shortages may worsen as AI adoption spreads, though the industry could remain cyclical; a reversal of the forced rollover trade is hoped for, not assured. x.com  x.com  x.com 

Semiconductors may face deflationary telco-equipment economics as chip throughput improves and compute deployment eases. Nvidia's tokens per chip rise dramatically each year; three or four deflationary forces could converge. LLMs sell tokens as carriers sold minutes, leaving consumers the main beneficiaries after penetration. If that analogy holds, few bulls would like either business. App developers may benefit after the customer ramp, though LLM providers could build their own. x.com  x.com  … Show more

Semiconductors could capture scarce spot compute's value if its forward curve holds. Compute trades in deep contango, while chip stocks price in little of that value; near-term moves are unpredictable. That makes the sector very bullish. The earlier lab-funding risk looks less acute: Anthropic should hold nearly $200B cash after its IPO and is operating-cash-flow profitable. Even with 2029 capex cuts, chips look cheap if future buyers do not default. x.com  x.com  x.com  x.com  x.com 

AI agents' productivity over human labor should lift GPU demand and rates for neoclouds. The bottom looks in for CoreWeave, Nebius, Boost Run, Sharon AI and IREN. Gavin Baker says token prices can fall as revenue per GPU rises; paraphrasing Elon, even $38 per GPU-hour can pay off. CoreWeave's five-year fixed contracts may lag current rates; Boost Run has less capacity locked in. If SpaceX takes more growth, slower cash burn could rerate neoclouds, though smaller customers may pay less. x.com  … Show more

Tech and semiconductor bounces are mainly driven by less urgent hedge-fund selling rather than fresh buying. Per Goldman Sachs, tech is still being sold amid very low activity; there is no sign risk has been added yet. The unwind may be nearing its end, especially as hyperscalers stick to capex plans and Q3 semiconductor earnings look likely to be positive. Less crowded positioning and more normal multiples leave a healthier base for a longer-term rebuild. x.com 

AI usage outstripping compute supply for the foreseeable future supports semiconductor demand despite cheaper tokens. Token prices fell about 90% in roughly three months, yet usage rose by more than that drop, so compute pricing keeps rising. AI revenue extends beyond LLM providers, including Meta, Google, Salesforce and ServiceNow. On- and off-balance-sheet debt is a risk; an AI bubble is likely eventually, but none is visible now. x.com 

Semiconductor fundamentals have improved as SpaceX and Meta say spending will rise further, while rates have come down. Earlier caution rested on fears that megacaps could not sustain spending and rising rates would slow AI investment; neither worry has played out. Today's selling feels more like stops being hit and investors giving up. x.com  x.com 

Semiconductor trade faces a capex growth ceiling even if AI delivers what bulls expect. Spending cannot compound at 70-100% a year indefinitely, and compute per dollar roughly doubles every two years, so the same budget buys more capacity later. Today's chip valuations also demand years of 20% AI earnings growth beyond 2028: roughly 10.9 extra years for AMD, 8.9 for ASML, 7 for Marvell and 6.4 for TSMC, versus 1.6 for Broadcom and 0.8 for Nvidia; Micron needs none under the chart's assumptions. … Show more

Hyperscalers' forward P/E has contracted 20% even as estimates ramped, making it rare to see things get much worse. Rising GPU rents and new lows in availability make the AI fears a fade; hyperscalers should lead. The next two weeks will be gut-wrenching amid the OpenAI raise, Anthropic S-1, Fed, midterms and mutual-fund selling into year-end, but that is the time to buy. x.com