AkhenOsiris

Stock is up a shit ton...failed to participate...this stock sucks and everyone in it is dumb as fuck.

$SHOP Tobi on Sources Pod re: Agents: Tobi’s advice to AI startups is to focus on the value their agents create before settling on a business model. I pressed him on the assumption that agents will collect a cut of every transaction they handle, and whose margin that money comes out of? He sees several possibilities, including charging users directly, metering usage like electricity, and taking a share of transactions. “If you build something that’s better than a cup of coffee, you are going to probably endure,” he says. That comparison felt familiar: when I priced this newsletter, I was thinking in cups of coffee, too. Advertising adds another complication: an agent paid to steer your purchases can become “a breach of trust,” but he still sees an opportunity for a new kind of advertising format for agents that maybe works more like a discovery marketplace.

WSJ: The U.S. has told several Gulf Arab countries to be prepared for a short, intense military campaign after the U.S. midterm elections, people briefed on the conversations said. The communications through diplomatic and military channels are among a growing number of indications that the U.S. and Israel are readying to launch a new round of heavy combat operations against Iran, risking a major escalation in hopes of degrading Tehran’s strength and pushing it toward negotiations. The U.S. military is bringing in a Patriot system from Texas to beef up its missile defenses and has moved troops off bases where they were vulnerable in preparation for an escalation, officials familiar with those moves said.

$META starts working again tomorrow after close (IF $GOOGL misses on search 🫣)

Trivariate Research founder Adam Parker is still bullish on AI. “We can’t be sure how long the cycle will last, but most experts think this is a 10-year cycle, with datacenter build outs likely taking longer than that in some cases. Even if this ends up being a 7-8 year investment cycle, we are only halfway into it now, and we highly doubt selling the exposed securities this far in advance is prudent,” he wrote to clients. Parker noted that, while the AI trade began in earnest in early 2023, increased productivity from the technology hasn’t fully trickled into earnings yet. “Only 15% of companies were discussing any of these issues at the beginning of the year, and still less than 50% of the top 500 US equities are discussing any of them today. Cost-related productivity is IN FRONT of us, not behind us,” he said. Parker thinks the best is yet to come from the AI trade. “A legitimate bull case for the S&P500 is that earnings can stay above trend because of productivity on labor that is still not yet in the numbers,” he said.

Chinese models are dangerous

$TSM ctee.com.tw/news/20261027700… TSMC CoWoS reportedly raises price by up to 20%. Extended lead times and tight production capacity in packaging are impacting equipment manufacturing, prompting the entire supply chain to race to deliver and expand production.

David Solomon spent Thursday announcing a $500 million AI push alongside four of the companies building the most powerful models in the world. Then he told the room that most of the economy doesn’t need those models. “Frontier models and how frontier models develop is going to be a very, very interesting thing,” the Goldman Sachs CEO told the crowd at the launch of the firm’s new Small Business AI Institute at the bank’s headquarters in lower Manhattan, where Fortune was in attendance. “But there’s very little that we do where we need frontier models,” he added, seeming to speak generally of economic activity, in the context of small business. “The technology is moving very, very quickly, and the technology is going to continue to move quickly,” Solomon said. But “you have to separate at some level” the development of frontier models from the everyday tools businesses actually use, he added. Frontier models, he said, matter for cybersecurity, for defense and for health care, such as around drug discovery and molecule formation. The security fights and the “bigger picture frontier issues that are getting a lot of attention,” Solomon said, “are really separate from using the base tools that exist to support your businesses.”

OpenAI said On Tuesday a GPT-6.1 Sol agent working on an internal testing task on October 3 suppressed automatic ‌security ⁠alerts after deciding an earlier instruction not to send Slack messages also covered safety notifications. In a separate incident on October 6, a GPT-6 Astra agent removed a safety monitor while copying an existing piece of code and later incorrectly claimed it had ⁠not done so, OpenAI said. "In both cases, the agents considered whether monitoring safeguards should be preserved, but interpreted routine coding instructions as permission to weaken them," the company ⁠said. OpenAI said it was making it harder for models to disable monitoring, adding that it does not "yet have a satisfying explanation for this behavior". OpenAI said it was testing whether its models might have a tendency to avoid being monitored and look for reasons to justify doing so.

Some interesting data from Orosz (Pragmatic Engineer): Spending on AI agents is ballooning at all the tech companies I talk with this year. With 2027 coming around fast, many companies are seeking to put a realistic number on how much they really spend on AI. The change in PR throughput is a very simple metric, and one of those “readily-available” measurements at nearly all companies, It’s also one that probably has not been “gamed” by devs yet, and it’s possible to compare a “before AI” and “after AI” measurement with this metric. To find out more and to try and answer the reader’s question, I asked around and got to talk with engineering leaders and senior+ engineers at 27 workplaces. I also gathered aggregate data from CodeRabbit and Gitar. - Stats from CodeRabbit and Gitar. Weekly PRs by devs using the AI code review tool increased by around 2x compared to a year ago. Meanwhile, Gitar sees “AI-native” teams ship 60+ PRs/dev/week. - On AI spend: Spend peaked during the summer, and is falling. June and July saw peak monthly spend at nearly all companies, and costs have roughly halved for those on API plans. A mix of Opus 5.5 and moving to open models is the reason for the drop at most companies. - Subscriptions conceal concrete AI costs, but some subsidies are ending. Several companies mentioned that their subsidized AI subscriptions are coming to an end, so they need to prepare to either optimize costs, or have their bill increase 5-10x. - Claude and Opus 5.5 seem to be the favorite models. Within this rather limited data, Claude and Opus 5.5 dominated answers about where most of the spending goes. Cursor, Codex and GitHub Copilot were also mentioned. - Publicly traded SaaS company, ~2,500 staff, ~600 engineers, US+EU $1,200/month/dev is the budget limit, but a Staff Engineer at the company reports that fewer than 5% of devs hit this mark. AI has helped: triaging, research, POCs - Multinational company, 10K+ employees, US / EU / Asia $4,000/dev/month maximum capped spend. This is a hard cap. Engineers at this company use GitHub Copilot and can choose between the models. Right now, Opus 5.5 is the top choice for devs across the company, a current team lead told me. Velocity: 3x more stories closed than last year. This team lead has been using Scrum and Kanban for years at the company, which helps compare “business value delivered”, not just PRs.

$GOOGL Pichai on Big Technology Podcast Pursuing cutting-edge research is costly for Google, requiring compute and chips that otherwise could be sold at a high markup by its cloud business. But Pichai said allocating resources to frontier AI development is “ground zero” for Google. “Our cloud business is at any given moment in time forgoing short-term revenues because we are making compute trade-offs to support frontier AI development,” Pichai told me. Look, we are so committed to being at the frontier. It’s in our DNA. We won’t do it any other way. We’ve been thinking about the AI frontier for well over a decade now. It’s also clear to me that users, both in consumer and enterprise, will have access to some multi-agent framework for them, working all the time doing things. It feels like a moment of infinite possibilities. Remember, we are building our own compute, our own data centers. There are long lead times. The compute we have today is a function of decisions I made in 2024. Even if you think back to a year from today, last year the agentic coding takeoff hadn’t happened. The way I think about compute allocation is making sure Google DeepMind, both for research and frontier model development, is cutting edge in terms of compute access. After that, we plan for all our businesses. Re: Jassy comment on compute sold out until 2028 - If today a customer is coming out and asking us for compute in ’27, that is incredibly difficult to do. Even last year we went out and leased some short-term compute to accommodate some customers and help them out. But yes, we are all dealing with similar timelines. If your question is, can there be an overbuild? Yes, because you can have some constrained components. Remember, what matters is a set of everything you need to build out that compute. Land. Like 2,000 components or more. You could get constrained at some point. There could be things which cause bottlenecks, and then you could have overbuilds. There could be external shocks to the economy. There are macro risks. But am I confident that the compute we are planning for, we will have uses for it given enough time? Absolutely, yes.

$NET Cloudflare today announced that more than 10 million developers are building on Cloudflare, up from 7.4 million in the prior quarter.

Agent Hacking: The cybercriminal instead outsourced the hacking spree almost entirely to artificial intelligence, tasking AI agents ‌with spending hours autonomously probing and breaking into targeted sites, according to Snorre Fagerland, principal threat researcher at the cybersecurity company NorForge. Reuters could not ascertain the degree to which the breaches resulted in unauthorized extraction of data. Deere told Reuters it had taken "necessary remediation actions" following the incident and had "no indication that sensitive information was accessed." SharkNinja said the hack had hit a standalone site carrying only product support information and that the threat had been contained. Cummins said it was "investigating the report of a potential cyber-attack on a website operated by one of our third-party vendors." Their instructions were terse, sketching out their preferred intrusion tactics in fewer than 100 Chinese characters. As of late last week, their agents had attempted to break into more than 1,200 sites. Successful intrusions typically took around ‌23 hours ⁠of poking and prodding for vulnerabilities. Fagerland estimated that the campaign, which appeared to be using a mixture of China's DeepSeek and Anthropic's Claude Code, had chewed up roughly $5,000 worth of tokens, or about $50 per successful hack. Fagerland — who has worked in cybersecurity for nearly three decades — said he was impressed by some of the techniques the agents used to compromise their victims. In one case, after failing to find any weaknesses in a big company's public-facing application, the AI agents pored through digital records to find associated infrastructure, located a public-facing control panel, discovered a subtle weakness in the way it processed data, and after hours of trial and error, reported leveraging it to trick the server into executing malicious commands. "It's ⁠not straightforward to come up with solutions like that," Fagerland said. "It's not script kiddie stuff." Fagerland said he had previously assumed hacker-agents would need at least some supervision. He said it was now obvious that the technology had moved past that. "The development in this field is astounding," he said. reuters.com/legal/government…

$NFLX WSJ: The ousters of two of the company’s top creative executives are a sign that the streaming giant’s leaders feel the service needs to overhaul its content offerings. Co-chief executive Ted Sarandos had complained internally for months that Americans were spending less time watching Netflix and that the service was short on new hit shows, people familiar with the matter said. He also has told colleagues that he wants the company, which has experimented with everything from videogames to podcasts, to strengthen its core TV and movie offerings. As of the end of September, Netflix hadn’t added net subscribers any month this year in the U.S., according to analytics firm Antenna. Antenna tracks U.S. consumer purchases of subscriptions through sources such as credit card bills, but doesn’t include those bought through wholesale bundles. Netflix hasn’t disclosed subscriber numbers since early 2025. A Netflix spokeswoman disputed Antenna’s data, saying it is “not representative of Netflix’s business performance.” Netflix is on track to deliver on its previously stated full-year forecast, according to an internal memo announcing the layoffs seen by The Wall Street Journal. It aims to deliver 13% to 14% revenue growth and operating margins of 31.5% this year.

Jim Cramer 🤡 Club column: I believe that what happened last week we will find is an unwind of certain firms’ positions coupled with the collapse of some hedge fund we have never heard of that borrowed billions that we don’t know about from a firm that hasn’t disclosed its own losses yet for lending the manager the money. But I think that’s the case with CoreWeave, as it is the linchpin to the thesis that the buildout isn’t at its end. You either believe it, or you don’t. If you believe it, you are tempted to buy Monday morning as the group is ridiculously oversold. If you think it is all one big sham, then short away or sell. Bottom line I don’t think it’s a sham. Let me repeat that: I don’t think it’s a sham. CoreWeave is not Lucent. Nvidia is not Nortel, the almighty vendor of that era. We can step back and say this is just some skirmish in the giant war for or against the data center. Moreover, I think that both Anthropic and OpenAI will come public, and anyone who needs to get paid will get paid. And, I think that the states that ban the data centers will rue the day they turned down all the money that is now going to flow to the remaining states that haven’t voted or voted in favor. The deals the hyperscalers will offer are insane. Full circle: My view has many believers. Too many. Because obviously many of them borrowed money to make, not that investment, but that bet. They may not even understand the sophisticated chain that gets us to where we are now. They are, alas, the problem. Broken hedge funds, not broken paradigms. If I am right, then the morass clears up pretty quickly; the bigger hedge funds that had their positions sold out underneath them will be revealed. The price of CoreWeave’s debt goes down, as it was always juiced by credit default swap buying ; and even Oracle’s stock goes up. Don’t get complacent. The bears need to make their year, too. But be aware of how long Nvidia’s chips last.

😂 32% of Americans think Sam Altman is Republican and 22% believe he’s a Democrat. 29% said they don’t know who he is. Even a fake tech CEO was pegged as Republican. The POLITICO Poll tested the name of a fake tech CEO: “Marshall Ancoates.” Twenty-two percent of respondents said they believed “Ancoates” to be a Republican, while 18% thought he was a Democrat. 30% of Trump 2024 voters believe that “Ancoates” is a Republican, compared to 19% who believe he is a Democrat.

FT: AI labs and Big Tech groups have committed to spend trillions of dollars over the coming years on computing capacity, or “compute”. But those massive rewards can only be realised if a group of recently launched “neocloud” data centre providers first manage to finance the unprecedented construction effort. Higher interest rates and a souring bond market are now threatening their ambitions, and with them the AI build-out, said a number of people involved in the market. Higher borrowing costs and rising inflation have added to mounting questions about whether the revenues of leading labs such as Anthropic and OpenAI will be sufficient to cover the costs. Plans by a procession of new developers and cloud providers including SoftBank’s SB Energy and Nvidia-backed Firmus to go public have been delayed or mothballed, with OpenAI’s decision to delay its own initial public offering further damping sentiment. AI labs such as Anthropic and OpenAI have reasoned they are better off risking the loss of a portion of their investment by moving fast — even where they rely on inexperienced partners — than suffering a shortfall of computing power that prevents them delivering their models, according to people with knowledge of the matter.

$GOOGL $AVGO FT: Anthropic is offering Blackstone a high yield of more than 10 per cent to finance its latest chip deal, highlighting the start-up’s willingness to stomach steep borrowing costs to finance expensive AI infrastructure. The $60bn debt package, dubbed “Project North Star”, would help fund Anthropic’s procurement of about 1.5 gigawatts of custom semiconductors developed by Google, known as tensor processing units. It is expected to be the largest chip financing to date and a crucial test of investor appetite in AI debt at a time when lenders are increasingly nervous about tech companies’ heavy capital spending and borrowing. Blackstone funds would invest $9bn in the debt backed by Anthropic’s five-year lease payments on the chips, accounting for about half of the $18bn junior loans, according to people familiar with the matter. The private capital giant was offered a coupon of 10 per cent and an even higher total return after accounting for new-issue discounts and other fees, the people said. A borrower with a similar credit profile typically pays a yield of about 7 per cent in the public bond market, according to the ICE BofA BB US High Yield Index. The rest of the $9bn would be sold to a broader group of investors in the coming months, potentially with a lower yield, the people added. Blackstone declined to comment. Anthropic did not immediately respond to a request for comment. Blackstone told its investment partners that the latest chip deal would be a rare opportunity to earn high yield on Anthropic-linked debt, as it is expected to receive an investment-grade rating and vastly lower its borrowing costs once going public, the people added. Anthropic was assigned a one-notch-below-investment-grade rating by KBRA in June, and the latest junior debt is expected to receive something similar. The new debt package also includes about $42bn of senior debt guaranteed by Broadcom, which helps develop the chips with Google. When the capital has been drawn down in 2027, the debt could become tradeable in the institutional 144A bond market.

"So Scott, if we let air out of the AI trade, long end yields will come down, how is that?"