Sam

@wallstengine Oh and by the way, their satellites are way different - Starlink’s are low orbit routers, so they can launch 3169 in a year. $ASTS are more like space cell towers, which are orders of magnitude more complex. 🤦🏾‍♂️🤦🏾‍♂️🤦🏾‍♂️

@wallstengine The downgrade is right to point out timelines, capital intensity and execution risk. But it overstates Starlink’s relevance as a threat to $ASTS and underestimates how much their evolution disrupts terrestrial carriers, rather than wholesale satellite providers like $ASTS

Scotiabank Downgrades $ASTS to Sector Underperform from Sector Perform, PT $45.60 Analyst comments: "Without yet a single retail customer and faced with the challenge of orbiting ~50 satellites to achieve continuous service in a handful of markets in late 2026 or early 2027, ASTS’s share price at $97.60/share has once again overshot to what we see as irrational levels (market cap of $37B). Evidence of slow user adoption in the U.S. and Japan, modest ARPUs, and high capex (including duplicated satellites for new frequencies) means investors may have to wait until 2028 or 2029 for tangible equity free cashas flow (EFCF) generation. Meanwhile, Starlink’s accelerated fixed growth and global brand recognition means ASTS competes with a leader that already has, in terms of revenues, the equivalent of 340 million global direct-to-consumer (DTC) users (~680 million by the time ASTS launches in select markets). While it has been an ordeal for ASTS to launch seven satellites since 2017, in 2025 alone Starlink orbited 3,169 units. We disagree with those seeing ASTS as a meme stock; the technology remains highly disruptive and has potential for dual use. But we also disagree with those dismissing ASTS’s multi-year delays and Starlink’s unstoppable growth. The middle-ground scenario points to a valuation range between $45/share and $55/share. Sell." Analyst: Andres Coello

@stephenasmith Max is what SAS wishes he was. Max has had some terrible takes, but almost always provides an articulate, structured supporting argument. Meanwhile SAS is frantically looking through a thesaurus before filming to try and come up with big words to sound articulate…

My full response to Max Kellerman: youtu.be/iD5a6doMZpY?si…

@ChrisCamiller1 Efficient 2007 Melbourne Cup 🔥

Makybe Divas third Melbourne Cup for me

@Racing_Previews Now compare that to the absolute slaughter job by Jye McNeil on Valiant King….

Jmac somehow finding the rail and first third of the field (the golden position in the Cup) from barrier 22 on Meydaan before they even got to the post the first time… 🐐 stuff. Best losing ride in the race, horse just didn’t stay.

@caseohcrushme An absolutely horrible ride. Couldn’t believe what I was seeing. Made no attempt to get inside, on a horse that you know can settle well. Just needed to find the rail, which it should have done easily from barrier 10. Pathetic

Jye McNeil literally sat 4 wide on Valiant King for the ENTIRE race, never tried to do anything for 3200m unbelievably putrid. Could have chucked an apprentice on and got more. #MelbourneCup

@_J0EL_ Depends on tempo. Definitely a chance in a slow run race, as it has far better acceleration that most here. But will struggle to stay off a decent tempo, which is what I expect. Class horse no doubt though

@G1Select Can’t knock these picks. Feels like half yours will be right there but may struggle the last 200. Buckaroo will need a slow run race or will struggle to see it out. Very keen Presage and Valiant. OSO needs a dry track. Wouldn’t rule out Torranzino

My ratings and thoughts on the 2025 Melbourne Cup.

@7horseracing @nickcat7 Jeez what could have been for Aeliana if she doesn’t get trapped in

Bruce, Emma and Lizzie take you through the 2025 Cox Plate from above.

@TSHickman Shocking ride. Had an easy run for most of the race then caught asleep when everyone started moving.

@mikealfred Virginia currently leads because of access to fibre optic networks and sub sea cables, but as power becomes a more important part of the equation, Texas will be right there. Look at the growth in North Dakota…

@mikealfred This is a contrarian view?? Cheap energy, significant land resources, low latency connectivity to both coasts, skilled trade and population growth etc.

Texas will become the largest data center market in the world both in terms of power usage and total economic output. This is a completely contrarian view today but I will be proven correct in time. Mark it.

@Jason can you please discuss Fermi America’s Hypergrid project in the next pod? America’s first BTM vertically integrated A.I. project using nuclear. If this works, could be the start of a nuclear renaissance in the U.S., where China is way ahead @DavidSacks @chamath @friedberg

@RihardJarc Neoclouds optimize for AI-specific workloads, where performance is king and multi-tenancy is less of a concern. They can run AI workloads in bare-metal Kubernetes clusters, avoiding hypervisors and maximizing raw performance

@RihardJarc You’re missing a technical detail. $MSFT $GOOGL $AMZN optimize for general-purpose computing, which requires hypervisors for security, multi-tenancy, and compatibility. This introduces inefficiencies.

There are multiple reasons why it makes sense for hyperscalers like $MSFT, $GOOGL, and $AMZN to work with neoclouds right now: 1. Demand for AI compute is much higher than the supply, so by working with neoclouds, hyperscalers can still make their clients happy while maintaining the front relationship and future client demand. 2. Hyperscalers are already spending enormous CapEx numbers on Data center compute and data center build-outs, and are limited on how much more they can spend. At the same time, you have to not overly spend in one year on one generation of GPUs, as each year a new generation is much better than the old one. To reduce that risk, you would rather work with neoclouds in the short term, where they take the hardware risk while you still maintain the client relationship. 3. Hyperscalers, because of their volume and "brand effect" (just look at the stock price moves of Neoclouds once they announce hyperscaler deals), can get good unit economics from Neoclouds, where Neoclouds don't have much margin. 4. Once AI compute demand settles down or data center build-out comes to a more normal time, hyperscalers can cancel their deals with neoclouds and switch that volume back to their own proprietary data centers.