Netflix Hits Record Revenue as SpaceX Turns First Profit and AI Infrastructure Debate Intensifies

Key Takeaways

1. Netflix posted record quarterly revenue of $12.9B, up 12% year over year, with net income up 44% and a record 34.6% operating margin, though growth was its slowest since Q3 2023 and the stock had fallen 45% over the prior year.
2. SpaceX reported $27.1B in revenue for the first nine months of 2026, up 98%, with net income positive for the first time at $4.4B, alongside $48.1B in capex and negative free cash flow of $35B year to date.
3. Raoul Pal argued AI infrastructure buildout is being underestimated, with compute arriving by 2027 and power, robotics and new chips scaling by 2028–2029, leading to an "economic singularity" in which intelligence and labor become extremely cheap.
4. Mirendil, in a post shared by Marc Andreessen, claimed it completed large-scale pretraining without a pretraining team in under four months at a compute cost below $10M, targeting recursive self-improvement.
5. US housing data showed September pending home sales at a record-low 66.0 index reading, down 50% from the pandemic peak and 35% below 2019 levels, with sales down 11% year over year.
6. Ben Carlson noted S&P 500 annualized returns of 22.9% over 3 years and 11.9% over 50 years, raising the question of whether investors should expect lower future returns.

1. Company Earnings and Capital Spending

1. Netflix reported record Q3 revenue of $12.9B, up 12% year over year, marking its slowest growth since Q3 2023. Net income rose 44% to $3.7B, and the operating margin reached a record 34.6%; the stock had fallen 45% over the prior year before the report 1 2.
2. SpaceX posted $27.1B in revenue for the first nine months of 2026, up 98%, with net income positive for the first time at $4.4B. Capital expenditures rose 247% to $48.1B, and free cash flow was -$35B year to date 1 2.

2. AI Infrastructure and Model Development

1. Raoul Pal said the pace of AI infrastructure buildout is underestimated: compute will be built by 2027, and power, robotics and new chips will scale by 2028–2029, at which point "everything collapses in price" as intelligence and labor become extremely cheap and old economic rules stop applying, a state he calls the "economic singularity." He described the window before it as "extremely short" and the best opportunity most people will see in their lifetimes 1.
2. A post shared by Marc Andreessen described Mirendil, founded less than four months earlier, as claiming it completed large-scale pretraining without a pretraining team, targeting recursive self-improvement to cut the time, human experts and compute needed for frontier AI. The post said Mirendil-Base was completed by one researcher in 3.5 months at a compute cost below $10M, with model tiers from 100B–500B to 2T–10T parameters, and argued a single researcher could do the work of a whole research organization, shifting the bottleneck to the ambition and judgment of the person directing it; independent evaluation, post-training benchmarks and API access were to be announced later 1 2.
3. Ben Carlson shared a view that most forecasts about AI's future ignore the frictions that exist in the real world, and separately raised the question of whether an AI bubble is bursting; his "Animal Spirits" segment listed whether CoreWeave is a canary in the coal mine and whether the bond market matters more than AI 1 2 3.

3. Markets, Housing and Investor Positioning

1. US pending home sales fell to a record-low 66.0 index reading in September, down 50% from the pandemic peak and 35% below 2019 levels, worse than any point in the 2008–09 downturn, with sales down 11% year over year; the shared post attributed the decline to high prices and mortgage rates back above 7% and called it the largest homebuyer recession on record 1. Separately, September single-family and multifamily housing starts fell 4.6% and 7.2%, and pending home sales growth slowed to -12.7% from -2.9% the prior month, the weakest in three years 2 3 4.
2. Ben Carlson reported S&P 500 annualized returns of +22.9% over 3 years, +16.2% over 7 years, +15.3% over 10 years, +11.2% over 20 years, +10.3% over 30 years and +11.9% over 50 years through September 30, and asked whether it is finally time to expect lower future returns 1.
3. Raoul Pal said he bought only twice during major market pullbacks over the past year and added to ZEC, judging undervaluation by an asset's deviation from its long-term log trend rather than trading cycles, as when he accumulated Solana and Ethereum in the 2022 decline. He acknowledged that global liquidity has been less correlated with crypto prices than expected, with much capital flowing to AI tech stocks, but said that could reverse; he called Ethereum and Solana established infrastructure networks and predicted a crypto-and-tech supercycle lasting to 2030, similar to the late-1990s boom but with severe pullbacks along the way. He said his publicly supported SUI position is underwater at an estimated average cost of about $1.70 excluding foundation tokens, and that adoption is the key test for his thesis 1.

4. Disputed and Unverified Claims

1. Spruce Point Capital questioned whether IperionX's ($IPX) titanium sales opportunity has gone to zero after Apple announced a partnership with Perryman Co to supply titanium products with 80% recycled content, given IperionX's prior promotion of consumer-electronics applications 1.
2. A post shared by Jason cited a research paper finding that ChatGPT secretly builds psychological profiles of users and that 96% of stored memory content was saved without the user's request; Jason commented that unregulated AI agent behavior is a black box, called this unsurprising, and said the industry should make safety and privacy top priorities. The claim is disputed/unverified 1.
3. Jason also raised the idea of releasing an open-source, working version of the Theranos vision, even at 10% of the original ambition, and predicted some would later say Elizabeth Holmes was simply ahead of her time 1.

5. Policy, Philanthropy and Public Debate

1. Meb Faber and Clifford Asness both described Elon Musk as having made the largest charitable donation in history and criticized Kara Swisher's reaction calling it "ridiculous waste of money," "only one share" and "I blame him for it"; Faber called the response predictable but regrettable, and Asness called it pathetic 1 2.
2. Meb Faber shared and endorsed the view that public pensions should have zero allocation to private equity, saying "fire them all, buy some ETFs," and noted that outside managers could become future employers, making pension negotiations awkward 1.
3. Liz Ann Sonders cited a Philadelphia Fed services survey in which the prices-paid component rose 4 points to its highest since April while prices-received fell 11 points to its lowest since April; a special question found 44% of non-manufacturing firms expect to increase capex in 2027, mainly on software, non-computer equipment and computer hardware, while a quarter plan cuts 1 2. She also noted that Japan remained the largest foreign holder of US Treasuries in August but reduced its holdings for a fourth straight month, while UK holdings exceeded $1 trillion for the first time 3.
4. Jason supported a mechanism to report government program fraud and collect a reward, saying the NDS team is advancing it, and questioned whether Anthropic's 5% public-interest fund could become a "slush fund," arguing the money should go directly to the public as shares and asking whether AI gains should be controlled by labs, the government or all Americans 1 2.

— via 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27