- The Markup
- Posts
- Ballots, Blocks, & Bedrock
Ballots, Blocks, & Bedrock
Anthropic is writing itself a new set of voting rules. OpenAI finally named the year. SpaceX hit 100 launches before the end of August. Kalshi got itself a Wall Street block desk. xAI landed inside AWS procurement. Perplexity found out its India giveaway actually converted. Anduril is protesting a contract it partly won.
Let's get into it.
π Portfolio Pulse
π§ Anthropic: Writing Its Own Voting Rules
Q2 revenue topped $11.5B, up from $787M in the same quarter last year, and the company posted positive adjusted operating income, which no other frontier lab can claim. Q1 was $4.73B, so it more than doubled in three months. Ahead of a listing that could land this fall, Anthropic is preparing a supervoting share class for Dario Amodei and six co-founders who together hold under 5%. The pre-IPO revolver is set to clear its $10B target, with lead banks asked for roughly $1.25B each. The figures are preliminary. The control question is the one buyers will actually price.
π€ OpenAI: A Date, a Loss, & an Ad Business
CFO Sarah Friar told an all-hands that OpenAI "will be a public company in 2027," or sooner if the business keeps inflecting, with slides putting the revenue run rate up 35% quarter to date and enterprise up 50%. The trailing picture is harder: Q2 revenue came in at $6.7B, up 18% sequentially, while operating loss widened to $12.3B from $9.3B. And on Aug 24, ChatGPT Ads goes live in 31 European markets, its largest expansion yet, on Free and Go plans only. Second revenue engine, first hard deadline.
π SpaceX: 100 Launches Before September
The 100th orbital mission of 2026 flew from Vandenberg early Wednesday, a Falcon 9 carrying 24 Starlink satellites, with booster B1097 landing on Of Course I Still Love You. That is the third straight year above the century mark, and it arrived roughly two months earlier than the same milestone in 2024, the year SpaceX finished at 138. Cadence is the input to everything else: Starlink capacity, national security throughput, and the revenue line public shareholders now underwrite. The rocket is no longer the hard part.
π Kalshi: Wall Street Gets a Block Desk
Cantor Fitzgerald opened institutional block trading in Kalshi event contracts, acting as introducing broker with Susquehanna providing pricing and liquidity. Trades get negotiated at a single price away from the central order book, which is the specific thing keeping roughly 3,000 institutional clients out until now. Cantor is one of the first full-service banks to do this on a CFTC-regulated venue. Meanwhile Nevada is back in court seeking $120,000 a day, saying its investigators placed nine trades after the geofencing deadline. Institutional demand on one side, state enforcement on the other.
π§ xAI: Grok 4.6 Lands in AWS Procurement
Grok 4.6 went generally available on Amazon Bedrock, a week after xAI shipped the model. It brings a 500K context window and configurable reasoning effort, tuned for long-running agents and coding work. The benchmark is not the point. The channel is: enterprises that mandate inference through a single cloud can now buy Grok the same way they buy everything else. A consumer-and-X-native lab does not build durable enterprise API revenue any other way.
π Perplexity: The India Giveaway Converted
Combined in-app purchase and subscription revenue in India rose about 60% from February through mid-August, measured against the stretch when the free Airtel offer was still open to new users. Downloads over roughly that same window collapsed more than 90%, to 3.3M. Monthly actives sat near 14M in July, down 37% from the October peak but more than five times the 2.6M average before the Airtel deal existed. Net mobile revenue reached $156,000 in July. Small absolute numbers. The cohort stuck, which was the entire question.
π‘οΈ Anduril: Protesting a Contract It Won
DHS split a five-year, $1.5B counter-UAS vehicle across 12 companies, one track for hardware and software, one for comprehensive services. Anduril landed the services track and missed the hardware track, the one covering products like Anvil and Sentry. It filed a GAO protest challenging the evaluation, with a decision due by Nov 20. Track one went to BAE, CACI, Leidos and four others. Winning the services half of a counter-drone contract is not what Anduril builds hardware for.
π Watch List
π³ Stripe: Buying the Meter on AI
Stripe agreed to buy OpenRouter, the gateway that lets developers route requests across hundreds of models from dozens of providers, for a reported $7.5B. OpenRouter had raised at roughly $1.3B a few months earlier. Stripe already moves money for a large share of the AI economy; OpenRouter puts it on the cost side too, owning the routing and metering layer between the apps and the labs. Model choice is becoming a procurement decision, and procurement decisions need a meter. Stripe just bought one.
π‘οΈ Shield AI: Navy In, Air Force Out
The Navy and DIU put $50M into X-BAT, Shield AI's autonomous VTOL strike jet, under the Runway Independent Maritime Expeditionary Strike effort. The Air Force is publicly holding off, which matters because X-BAT was positioned squarely against the Air Force CCA program. Shield AI puts the unit price near $27M, right at the CCA target. The aircraft has not flown yet. One service sponsor plus an unflown airframe concentrates the risk in exactly two places: Navy budget lines and first flight.
π¬ Prompt
What can I help you with?
Iβm about to buy this product on Amazon for [PRICE]: [PRODUCT LINK OR NAME].
Check whether this price is genuinely discounted or if the seller likely inflated the original price before running a fake sale. Use free tools to analyze the productβs price history, show me the normal price range, the lowest recent price, how often it goes on sale, and whether the current deal is actually worth buying. Also tell me if I should buy now or wait for a better deal based on historical pricing trends and upcoming seasonal sales.

π Content
βοΈ Kalanick on Chaos, Lawfare, & Atoms
Travis Kalanick runs the whole arc: Red Swoosh, Uber's hyper-growth, and the China campaign where the actual job was finding the line between order and chaos. His frame is "problem solver in chief," meaning you predict what kind of problem is coming next and staff your management capacity ahead of it rather than reacting once it lands. The VC section is the one founders should sit with: build a process that manufactures alternatives instead of signing the first term sheet. On leaving Uber he is blunt that lawfare and a turned media narrative did the work, and that business now runs like politics. Then he gets to Atoms, digitizing the physical world with AI and purpose-built robotics across food, mining, and transportation. Must watch.
π¨οΈ Your Read-Later Pile, On Paper
Daniel Belfort launched Offprint, a physical magazine printed from everything you save and never actually go back to read. Text the service anything read-worthy all month, approve the issue with one reply on the 1st, and it lands at your door a week later for $19. Find online, read offline. He shipped it this week and traffic was heavy enough that he bought and deployed a second line to keep up. The most contrarian product of the week is a paper magazine, and we are going to try it out.
πΈοΈ Automating the Luck Surface Area
Ed Leon Klinger pointed an AI at his own corpus: every meeting, email, Slack, LinkedIn message, and thousands of call notes. It scores relationships on strength, recency, communication style, and likely mutual value, then proposes the intros worth making. Left running overnight, it did its own desk research on his contacts unprompted, flagged people who had given him intros and advice without getting anything back, and matched a colleague's year-old description of an ideal mentor against someone in the graph. This is the part of relationship work that never scaled. Seeing a connection and making a connection are two different things.
π€ Get in the Room
Private investing for (almost) everyone*
We are a curated network of exceptional angels, VCs, family offices, founders, and industry experts investing in pre-IPO companies. We back sharp operators and high-leverage deals. We move fast. We bring value. We don't miss around the edges. Deals don't wait. Neither do we.
*Vibes and values must be aligned. Must be an "accredited investor."
Stay cool,


Β© 2026 LFG Ventures
All rights reserved.
The content presented in The Markup is for informational purposes only and does not constitute an offer or solicitation by LFG Ventures to buy or sell any securities. Nothing in this publication should be interpreted as investment advice, a recommendation, or an offer to provide brokerage services. Any offers or solicitations will be made only through official offering documents and subject to their specific terms, risks, and conditions.