{"slug": "election-jitters-self-improving-ai-and-wall-street-tantrums", "title": "Election jitters, self-improving AI, and Wall Street tantrums", "summary": "Anthropic's IPO filing states that its internal estimates and expectations for future models, rather than its current models, indicate the company may imminently meet its Responsible Scaling Policy's definition of \"dramatically accelerated progress attributable\" to AI systems building their successors, according to the S-1 filed with the SEC. The filing also covers token prices and the effect of falling AI prices on AI labs, while ChatGPT reached 1.2 billion weekly users and Palantir and Shopify fell after reporting earnings. The newsletter's author, Alex, expects Democrats to underperform expectations but capture at least one chamber of Congress in today's US midterm elections, with Kalshi giving Democrats a 77% chance of winning the Senate and 94% of the House.", "body_md": "# Election jitters, self-improving AI, and Wall Street tantrums\n\nWhat, me worry?\n\n**Tuesday.** America votes today; Palantir and Shopify got whacked after reporting earnings; and Anthropic gave us a fascinating window into just how much it costs to release a lower-cost model. To work! **Your friend, Alex.**\n\n- ***📈 Trending Up:*** [*** Well, that's embarrassing***](https://cloud.google.com/blog/topics/threat-intelligence/shutting-the-door-on-the-xcpe-botnet/) ***...*** [*** editing weights***](https://rikiyat.github.io/blog/weight-space-programming/) ***...*** [*** Salesforce's M&A team's Red Bull budget***](https://www.theinformation.com/briefings/exclusive-salesforce-acquires-startup-cresta-2-7-billion) ***...*** [*** hacking on accident***](https://x.com/joshmanders/status/2117458724407415295)***? ...*** [*** Lattice, after earnings***](https://ir.latticesemi.com/static-files/9078e104-bc26-4f73-ae39-a85e95219c37) ***...*** [*** running K3 at home***](https://github.com/FareedKhan-dev/kimi-k3-in-c)***? ...*** [*** ChatGPT, now with 1.2 billion weekly users***](https://www.bigtechnology.com/p/openai-chatgpt-search-is-about-20) ***...*** \n- ***📉 Trending Down:*** [*Keeping your crypto*](https://www.theblock.co/news/defi/2026-11-02-mercado-bitcoin-hack-46-million-brazil-exchange-432333) *...* [*Apple, after earnings*](https://www.cnbc.com/2026/11/02/apple-aapl-q4-2026-earnings-live-updates.html) *...* [*democracy*](https://x.com/AdImpact_Pol/status/2117632154196652068) *...* [*US-Israel ties*](https://www.newsnationnow.com/politics/ap-politics/ap-trump-says-us-will-halt-4b-in-aid-to-israel-as-country-shifts-to-buying-military-hardware-instead/)*? ...* [*human-designed chips*](https://www.theinformation.com/articles/anthropic-touts-claudes-chip-design-skills-ipo-investors) *...* \n\n**Election Day:** Today's the day! After this evening, you won't have to read more about how America will vote. Instead, you will be treated to weeks of discussion concerning how America *did* vote. Lucky you.\n\n- [Kalshi](https://kalshi.com/category/elections/midterms) gives Democrats a 77% chance of winning control of the Senate (as of the time of writing), and a 94% chance of Democrats winning the House.[Polymarket](https://polymarket.com/midterms) gives Democrats a 78% chance of winning control of the Senate, and a 95% chance of winning the House. Polymarket traders also give the opposition party a 91% chance of \"holding the most governorships\" after the polling.\n- Lots of tech-right folks think POTUS and his party will do better than expected; most left-tech folks I see are assiduously avoiding counting unhatched chickens. (The bespoke concern to have is what will happen to [prediction market volumes](https://dune.com/datadashboards/prediction-markets) after the election concludes!)\n\nMore than control of Congress is at play. A host of local votes on data-center-related issues are planned, including plebiscites [in cities in Illinois, Maine, and Ohio](https://ballotpedia.org/Data_center-related_ballot_measures,_2026). The controversial — and we consider self-defeating — wealth tax in California is also on the ballot.\n\nAs discussed yesterday, I expect Democrats to underperform expectations but capture at least one chamber in Congress. I would be happy to be found too pessimistic. Let's see.\n\n## RSI, token prices, and revenue growth\n\nA few final notes on the latest Anthropic IPO filing. Let's talk progress towards recursive self-improvement (RSI), token prices, and what impact falling AI prices have on AI labs in particular.\n\n**RSI cometh?** Here's Anthropic [discussing](https://www.sec.gov/Archives/edgar/data/2130357/000162828026068676/anthropicpbc-sx1ax1.htm) how close it is to its AI systems building their successors:\n\n[O]ur internal estimates and our expectations regarding future models, rather than our current models, indicate that we may imminently meet our Responsible Scaling Policy’s definition of “dramatically accelerated progress attributable to AI.” [...]\n\nYou can read Anthropic's scaling policy document [here](https://cdn.sanity.io/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf) (page 8) for more. But the idea that Anthropic is *close* to RSI is important. Anthropic is already the world's leading AI model maker; if it reaches RSI first, and the result of reaching the threshold accelerates its work further, we could see the company pull ahead of its rivals at an even faster pace than previously seen.\n\n**More intelligence + lower cost = more usage = faster revenue growth?**\n\nEarlier in 2026, Anthropic found itself in the midst of a storm. After rapid adoption of early agentic AI tools (largely coding-focused), corporations discovered that the short-lived tokenmaxxing era had to end. Expensive AI models suddenly lost some of their luster. Since the AI cost crisis, OpenAI has made great strides in offering frontier intelligence at competitive price points. Anthropic is getting in on the fun. Here's the company discussing the recent release of Opus 5.5 (emphasis added):\n\n **[W]ith the release of Opus 5.5 that same month, we reduced customers’ cost per task by approximately 40% compared to Opus 5.** In each case, the reduction was a combination of a lower price and improved efficiency, the contributions of which depend on the task. Because of our compute efficiency, we have been able to make these improvements while preserving our margins.\n\nSelling a model that is better than its predecessor at a lower price point? Sounds like business suicide. But Anthropic has found that offering more intelligence for less money is a great way to drive adoption (emphasis added):\n\n**In the past, when we have launched new model versions that deliver a significant increase in capability and a lower cost per task, we have generally seen customers expand their use of our intelligence service.** For example, we launched Opus 4.5 in November 2025 with greater capabilities and a lower price and cost per task than its predecessor, Opus 4.1. Following that launch, our total quarter-over-quarter revenue growth accelerated from 48% for the three months ended December 31, 2025 to 132% for the three months ended March 31, 2026. **This growth reflected the combined effect of several factors. These included lower prices and improved capability that unlocked new use cases, the emergence of new products, and expanded customer adoption.** \n\nEven more, Anthropic writes that after releasing the more performant and cheaper Opus 5.5, \"many customers have since migrated to it from older models.\"\n\nBoy did they. Anthropic's reduced model prices and certain contractual discounts came with a material tradeoff (emphasis added):\n\nFor the three months ended June 30, 2026 to the three months ended September 30, 2026, **the increase in revenue due to higher usage was partially offset by a decrease of approximately $600 million from price reductions introduced as part of new model versions** and contractual discounts. Price changes did not have a material impact on the other quarterly revenue trends presented.\n\nThere's a real cost, then, to offering lower-priced models to customers currently using their less efficient ancestors. Anthropic seems to be betting on Jevons' Paradox rather heavily, in effect happily taking the tradeoff of selling more for less in hopes of capturing share of a future, enlarged market. A bet I like to *think* I would have the guts to make. \n\n### Shopify and Palantir get whacked\n\nCanadian ecommerce titan Shopify reported a [rare growth miss this morning](https://www.investors.com/news/technology/shopify-stock-shop-shopify-earnings-news-q32026/). For the first time in years, Shopify's reported [revenue of $3.74 billion](https://www.shopify.com/investors/press-releases/shopify-sustains-rapid-growth-revenue-32-and-free-cash-flow) (+32%) fell short of street expectations of $3.75 billion; platform sales (GMV) also missed the mark, coming in at $119.145 billion (+29%) against expectations of $120.08 billion. (Subscription revenue also came in light: $836 million versus an anticipated $850.2 million.)\n\nShares are off 9% this morning, bouncing off morning lows.\n\nShopify's free cash flow margin rose to an impressive 22% in the quarter. Solid growth, strong platform growth, and lots of operating cash generation? What's not to like? Apparently, guidance, though the company's disclosed Q4 2026 expectations for revenue growth topped expectations. Best I can tell, Shopify's fourth-quarter guidance implies gross margin erosion, which investors don't like to see. That and Shopify sports a price/earnings ratio north of 60x. A high bar to clear, even for a company as healthy as the Great Canadian Hope.\n\nPalantir is another puzzler. The AI and data company [handily beat revenue expectations](https://finance.yahoo.com/markets/stocks/article/palantir-beats-q3-estimates-raises-full-year-revenue-outlook-as-us-commercial-sales-top-us-government-revenue-for-the-first-time-234245540.html) in the third quarter, [reporting](https://investors.palantir.com/news-details/2026/Palantir-Reports-Q3-2026-U-S--Comm-Revenue-Growth-of-148-YY-and-Revenue-Growth-of-96-YY-Raises-FY-2026-Revenue-Guidance-to-91-YY-Growth-and-U-S--Comm-Revenue-Guidance-to-143-YY-Crushing-Consensus-Expectations/) a top line of $2.31 billion (+96%), ahead of the $2.17 billion investors expected. Palantir's new full-year guidance easily clears street estimates, and it even beat third-quarter per-share profit expectations! So, why are Palantir shares down 4.6% this morning? The company's Q4 2026 forecast represents only about 15% growth vs. Q3 2026, less than the 19% it managed last year.\n\n- Yes, but companies tend to slow down as they scale, and with more than $8.5 billion worth of revenue forecast for 2026, Palantir is hardly a small company!\n- With both Shopify and Palantir, the best explanation for their trading results I can summon is that, with their relatively expensive valuations and AI jitters making tech investors nervous, you have to do better than ever simply to stand still.\n\n**How about some good news?**\n\nWhile the world bites its fingernails regarding the financial health of hyperscalers and growth at other trillion-dollar giants, the software as a service (SaaS) market has been on a bit of a tear recently.\n\nHere's the Bessemer cloud index, through the lens of an ETF that tracks its performance:\n\nFrom a 52-week low of 23.90, $WCLD is now north of 43. Not bad. You could even say *very good*. Indeed, SaaS shares are trading at their highest levels since 2022. And we all thought SaaS was dead! Even better, revenue multiples for public SaaS companies are recovering, though they remain depressed compared to their pandemic-era peaks.\n\nYou could argue that as AI loses some of its magic pixie dust, supposed victims of said dust *should* rise in response. Yes, that sounds reasonable. But we're talking about the stock market.\n\n*Alright. Enough. Let's go refresh Twitter all day and wait to see how the votes tally up.*", "url": "https://wpnews.pro/news/election-jitters-self-improving-ai-and-wall-street-tantrums", "canonical_source": "https://www.cautiousoptimism.news/oh-crap-its-election-day", "published_at": "2026-11-03 16:20:50+00:00", "updated_at": "2026-11-03 16:50:32.858849+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-safety", "ai-policy", "large-language-models"], "entities": ["Anthropic", "Claude", "OpenAI", "ChatGPT", "Palantir", "Shopify", "Kalshi", "Polymarket"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/election-jitters-self-improving-ai-and-wall-street-tantrums", "markdown": "https://wpnews.pro/news/election-jitters-self-improving-ai-and-wall-street-tantrums.md", "text": "https://wpnews.pro/news/election-jitters-self-improving-ai-and-wall-street-tantrums.txt", "jsonld": "https://wpnews.pro/news/election-jitters-self-improving-ai-and-wall-street-tantrums.jsonld"}}