He is running 7x -8x levered SaaS businesses. What else he could have potentially said?

$NVR - never hosts earnings call and never provides guidance. But things were so weak in housing in late 2025 that even they have to come out and caution investor about the weak demand environment they have been witnessing.

Frontier labs are on a treadmill to build ever bigger AI models so that they don't get leapfrogged. Only way to do this is by raising billions. Billions can be raised by talking about trillions in revenues. Not an AI sceptic, I am sceptic on rev numbers talked here.

Pretty amazing, many builders are +15% YTD. Were big beneficiaries of rotation out of tech and the hope that Trump Admin. will do something to get rates down. Fundamentals are marginally better with lower new home inventory, but not so great that $LEN, $DHI and $PHM to rip 15%.

$GOOGL's TPU and AI advantage: Gemini 3.1 is 50% more efficient relative to GPT 5.2 and Opus 4.6 and has 2x more reasoning capabilities than Gemini Pro.

Bernstein put out a framework to evaluate disruption risk to software companies from AI. Framework uses Automatability and Defensibility. Seems like market is only focusing on automation risk right now and does not give any credit to data moats. $CRM, $MDB, $NOW, $TEAM, $ADBE

We are in a market that shoots first and reasons later. It is tough to disprove any appocalyptic AI narrative. We need a few software companies to come out and post accelerating growth due to AI. This needs to happen for a few Qs before market gains conviction. $MDB, $SNOW

While Fintwit is excited about dreaming up AI doomsday, $OAI is reducing its compute spend from $1.4tn to $600bn. Even if AI adoption is massive in coming 3 years, we may run into compute shortages. $MSFT, $ORCL

All these $DASH advantages in AI era applies to #Eternal #Zomato as well. #Eternal is an expensive stock, therefore, there is little valuation support and can derate further. But the business will be fine.

We are in a market that shoots first and reasons later. 6% decline in $DASH is an indication of market raising odds of disruption risk. Ben Thompson put out a compelling case on $DASH not being an AI loser. An expensive stock can derate even if biz is going to be fine.

Anthropic hosting a presentation on Enterprise Agents and messaging coming out of #Anthropic is that they are working with software providers. Not replacing them. Not sure if the moves in software names stick. $CRM, $MDB, $TEAM, $WDAY

Continue to believe that market is not giving any credit to data moats and just focusing on what can be automated. Case in point, #Anthropic calling out $TRI as the only place to get Westlaw data. If you have data, then you will be more insulated. $MDB, $VEEVA, $TRI, $SNOW

$BKNG also acts as merchant of record. $GOOGL never wanted to act as merchant of record even when it launched its travel offerings in search. Same will be the case with other LLMs as it brings you under the preview of regulators. AI will take the path of least resistance.

Fail to understand what $CSGP is trying to achieve here. There are concerns around Homes.com gowth + AI disruption risk and market share loss for Apartments.com. By combining these two business segments, you are not helping investors to gauge health of biz

Buying $NVDA at current valuation has given 150% return with no instances of drawdown. Only thing that seems different is that buyside has already modelled in $500bn of backlogs. So numbers are not moving higher and the fear we may hit peak in 2027/28 has caused this derating.

$META's revenue per employee after recent layoffs now stand at $2.5 million+ vs sub $2 million in 2019. Fastest growth among big tech. AI should lead to massive cost tailwinds for most large cap tech names. $META, $GOOGL, $MSFT

$NVDA just raised its revenue guide by ~20%+ and stock is up barely 2.5%. The fear of a peak just keeps hitting the stock. Street estimates for 2026 and 2027 is ~$750bn and Jensen just said $1tn in revenues.

$TSM raising capex guidance due to N3 node capacity expansion and not due to bleeding edge is rather abnormal. Shows that there are a lot of GPUs/ASICs that need to migrate from N4/N5 node to N3 and repurposing N4/N5 tools to N3 is not along going to meet the demand.