Google’s AI search squeeze rattles investors

Startups (1)

• A Pirates Labs contact said a pure graphene sheet thicker than a millimeter would be extraordinarily valuable; another speaker claimed access to a major graphene manufacturer but could not recall its name.

BTC Markets (1)

• Salem called Chamath’s claim that Bitcoin is capped, has a limited buyer pool, and will eventually be hacked and go to zero a “scop.” Immad said AI and quantum computing could enable attacks, but Bitcoin could upgrade and retain buyers as a historic asset.

Custody (1)

• The panel described a $6.3 billion Binance theft, allegedly enabled by attackers linked to North Korea compromising one engineer’s machine; the funds were moved in 20 minutes, and about $3 billion was reportedly frozen. Tether and Circle could freeze some stolen stablecoins, while public transaction trails made the transfers traceable.

Banking (1)

• Immad compared the exchange breach with Bangladesh losing $100 million through a SWIFT transfer after an attacker impersonated a central-bank official. He warned that increasingly convincing AI-driven impersonation attacks could target both crypto users and traditional banks.

Media (1)

• The panel remembered Patrick Stewart, who the host said had died at 86, and praised his portrayal of Captain Jean-Luc Picard. Alex suggested the character would persist through repeated fan and AI recreations.

Models (3)

• Google’s Gemini 4 Carbon reportedly found 88% of vulnerabilities on SEC Bench Pro, exploited 55% of flaws on Exploit Bench, and wrote working patches for 71%. Alex said the model was not generally available and its benchmark position could reflect higher reasoning costs rather than a clear capability lead.
• Google’s Genie 4 generates interactive 3D environments with sound, simulated physics, and real-time prompt-based changes. The panel saw uses in games, robot training, factory simulation, construction planning, and visual management, while Peter estimated current operation at $20 an hour and predicted $2 an hour next year.
• Sam Altman said OpenAI spent hundreds of millions of dollars on mathematical research, with one proof costing around $10 million and later attempts costing less. Alex speculated that unpublished results could include Yang-Mills work, further Millennium problems, or cryptographic advances; panelists described the researchers’ excitement and sadness at a field-changing breakthrough.

Physics (1)

• Alex argued that physics research should make falsifiable predictions grounded in the physical world, criticizing work that does not attempt to do so. He said his early chain theory leaves the simulation hypothesis untestable, while Immad argued AI could weaken peer review’s gatekeeping and make scientific ideas compete more openly.

AI Infrastructure (1)

• Sam Altman said OpenAI is developing robots to build data centers, framing self-replicating infrastructure as a route to more compute. Alex argued lunar facilities could avoid terrestrial permitting resistance; Dave said data centers could make host regions wealthy, while the panel debated whether governments or AI systems would regulate them.

Robotics (2)

• Tesla said Optimus production is starting at Fremont, with a line designed for one million robots a year and first customer deliveries expected in 2027. The panel disputed how quickly demand would extend beyond factories, while Immad argued that robots could become valuable assets as software improves and Dave saw industrial construction as an early market.
• Immad argued that biological dogs retain advantages in endurance and adaptation, while mechanical dogs offer upgradeable software and repeatable production. The panel also discussed Elon Musk’s interest in longevity and Peter’s report of promising work at Isomorphic Labs, without detailing the underlying research.

Autonomous Vehicles (2)

• The panel compared Tesla’s projected Cybercab cost of 20 cents per mile with Waymo’s newer service at 40 cents and its older version at $16. Waymo reportedly had 4,000 driverless vehicles, 500,000 paid weekly rides, and service in 15 cities; speakers said chip supply, maintenance, and real-world operating costs will determine the contest.
• Cambridge Teamsters and transit workers asked local officials to block driverless taxis, citing jobs and safety. Panelists compared the resistance with Britain’s Red Flag Act and New York elevator-operator strikes, and argued autonomous vehicles could improve safety and lower travel costs despite labor disruption.

Education (1)

• Computer science enrollment fell, and US full-time MBA applications dropped after two years of growth; panelists said students fear AI will reduce entry-level jobs. Salem argued universities should become entrepreneurship institutes, while Dave and Peter emphasized building, networks, and purpose over credentials alone.

Brain (1)

• Peter said Fountain Life found that 45% of dementia cases are preventable through lifestyle, and that 46% of members with advanced brain age improved after 13 months of lifestyle support. The panel also discussed purpose after work disappears, while Peter cited Elon Musk’s growing interest in longevity and GLP-1 drugs.

Digital Sovereignty (1)

• Anduril and Qualcomm announced Blue Chip, a rugged US-designed computer for military, government, and emergency users. Palmer Luckey said about 90% of parts came from the US or allies, with a target of 100%; the laptop costs about 20% more than high-end PCs, and its screen remains made in China.

Enterprise (1)

• Salem said organizations should keep security rules, permissions, records, and compliance obligations above individual AI models so providers can be swapped without losing institutional accountability. He described machine-readable governance as necessary because models are changing every week or two.

China (1)

• Immad attributed lower Chinese AI prices to compute constraints that encouraged efficient model design, distillation, and strong inference engineering, alongside lower research costs and thinner margins. Dave warned that stacked margins at TSMC, Nvidia, and model providers may compress as companies design chips and build fabs themselves.

Labor (1)

• Alex interpreted Microsoft’s voluntary employee-exit offer as a way to shift cash from payroll operating expenses toward data-center capital spending, rather than straightforward AI substitution for software engineers. Dave expected sharp stock volatility for AI companies as each model release and benchmark changes investor expectations.

Religion (1)

• Salem argued that religions across traditions have used forced conversion and cited Christian inquisitions, Muslim persecution, Tibetan Buddhist suppression of Bon practitioners, and coercion under the Hasmonean kingdom. He said he intended to criticize all religions equally, then noted that roughly two-thirds of the world identifies as religious.

Startups (2)

• Jason says Anthropic is targeting a November 16 IPO at a $1.5 trillion to $2 trillion valuation, seeking $100 billion; the S-1 reports $16.3 billion in first-half 2026 revenue, up 14x year over year.
• Sacks says Anthropic's reported $80 billion loss came from convertible-note accounting, not cash burn; he describes the company as capital efficient, with improving margins and roughly 85% gross margin.

AI Infrastructure (1)

• Friedberg argues that AI companies and their customers can all earn strong returns: his illustrative stack estimates Nvidia turns each dollar of chip costs into four dollars of revenue, while Anthropic earns about 85% incremental gross margin.

Enterprise (2)

• Friedberg estimates Anthropic has reached only about 1% of its potential customer base for businesses able to spend $100,000 annually; Chamath sees revenue reaching hundreds of billions if larger customers prove ROI and expand spending.
• Chamath says he missed Q2 after pushing his sales team to focus only on huge deals; the hosts argue that smaller deals with customers who pay and start projects on time can build a large business.

Markets (1)

• Chamath sees AI infrastructure spending creating systemic financing risk: AI companies may compete with the U.S. Treasury for capital, pushing borrowing costs higher for either the AI sector or the wider economy.

Regulation (1)

• Friedberg predicts governments may regulate compute allocation as AI firms compete with governments for capital and influence; Sacks says Anthropic has built a large government-affairs operation and warns its IPO will expose its power as a major tech company.

Models (1)

• Sacks argues Anthropic's drive for recursive self-improvement helps explain its falling model costs, but creates tension with safety-focused critics; he says the company reported using RSI for a share of its model-development tasks in August.

Adoption (1)

• Jason says workplace AI use remains limited because companies restrict access to email and internal knowledge systems over cybersecurity concerns; he expects broader agent deployment once those risks fall, while Sacks says newer applications consume far more tokens.

Big Tech (1)

• Jason points to Microsoft growing revenue from $198 billion to $332 billion over four years while adding only 2,000 employees, arguing that AI can raise output per worker rather than simply eliminate jobs.

Elections (1)

• The hosts split on the midterm outlook: Chamath expects Republicans to keep the Senate and Democrats to narrowly win the House, while Jason predicts Democratic control of both chambers and Sacks expects a more typical midterm result.

Macro (1)

• Sacks rejects claims that the economy is a disaster and cites growth, income, and poverty figures; Jason counters that Trump's approval is at a career low and says inflation, war, and unmet promises have alienated some Republican voters.

Biology (1)

• Friedberg says farmers in Kenya were lining up to try true potato seed, a crop technology he encountered during his visit.

Enterprise (3)

• Weil, Gotshal & Manges filed for Chapter 11 after high-profile lawyers left, succession planning failed, and merger talks collapsed. The host says its bankruptcy is not primarily an AI-disruption story, despite pressure on large firms to adopt AI.
• Crosby submitted a letter of intent to buy Weil’s brand, trademark, and internal records, and hopes to hire some of its lawyers. Crosby is an AI-enhanced law firm with clients of its own, valued at $400 million and backed by Sequoia, Bain Capital, Lux Capital, and Index Ventures.
• Weil’s precedent documents, templates, and bankruptcy workflows could be valuable training material, although private client files are not part of the proposed sale. Harneet K. cited a Google bid for Spirit Airlines operational data; GenLegal CEO Ryan Walker argued client ownership makes Weil’s data unacquirable.

Big Tech (1)

• Amazon’s AI spending is being funded by a still-larger legacy business, while Intel’s foundry growth comes with heavy losses. CEO Andy Jassy expects AWS growth to stay above 40%, but the host notes AI businesses are not yet large enough to fund their own expansion.

AI & Tech (1)

• Democratic representatives Raskin and Garcia asked Trump whether he used Grok for official advice and demanded preservation of any chatbot logs. Their letter cited reporting that Trump sought Grok’s advice about removing Nicolás Maduro and warned that AI could affect decisions on Venezuela or Iran.

Models (1)

• Nathaniel Whittemore favors voice input for AI and says he now mostly speaks rather than types. Granola launched a free, unlimited dictation feature that types into any app on Mac, with iOS and Windows versions coming later.

China (1)

• The New York Times reported that US intelligence surveilled Megaspeed executive Alice Huang over suspected chip diversion and possible information-sharing with Beijing, including information about Jensen Huang. NVIDIA disputed the allegations, while Megaspeed said it hired outside counsel and would cooperate with an evidence-based investigation.

Agents (8)

• Spawn founder Jacob Sanbury built a gaming platform where players prompt an in-game agent to change a shared world. A 10-person team runs persistent agents with named identities, memory, and company functions; the fleet shipped 77,000 code changes in a month.
• Spawn spent $3.1 million on tokens in one month, 30 times payroll, and runs 5,000 CPUs, 17 terabytes of RAM, and 150 GPUs on AWS. Sanbury says Fable made the system functional and increased the number of changes the team could ship 500-fold.
• Spawn assigns agents system access and ownership, then uses helper agents for granular work while employees steer larger decisions with modified Claude Code. Sanbury says OpenAI models game checks, Google models lose coherence, xAI models lack intelligence, and Anthropic models work best but take too little risk.
• Spawn removed pull requests because agents understood the whole system and spent too much time coordinating. Agents work directly in the codebase, fix breakages as they arise, align the fleet hourly, and run test suites every 15 minutes; about 1,000 of 123,000 changes were reverted over a year.
• Spawn closes the feedback loop by having player agents report bugs and feature requests directly into the fleet, where a small model triages them. Sanbury described an issue detected eight minutes after affecting a player, diagnosed a minute later, and fixed 11 minutes after that.
• After Sanbury faced 700 requests, Spawn turned two years of human decisions into principles for AI judges. Judges reuse precedent, return unclear requests for revision, or escalate novel cases; they made over 11,000 decisions in eight days, leaving Sanbury to decide only 27.
• Spawn’s human engineers now act as rapid-response generalists: they shape product taste, delete work made redundant by new models, and push for ambitions agents would not propose. Sanbury scrapped six months of work covering 70% of the engine and says a prompt about a world 1,000 times GTA’s size led to Spawn’s latest build.
• Sanbury says agent-first operations require closing the production feedback loop and tolerating agents breaking things; he also warns against forcing agents into human-shaped workflows. Nathaniel Whittemore highlights persistent agent responsibility, shared work context, observability, decision precedent, and human direction as practices other teams can adapt.

Models (8)

• Gemini 4 Carbon scored 59 on Artificial Analysis Index v5, a 9-point gain, and led six of eight highlighted benchmarks. It gained 8 points on Terminal Bench 4.0, 6 on Frontier Suite, 12 on Exploit Gym, and 7 on OSWorld.
• Google released Gemini 4 Carbon two weeks after Argon but limited initial access to cyber defenders and trusted partners while it refined guardrails. Arena reviewer Peter Gostead said Carbon felt much better than Argon; Xiao Zhang described weeks of internal use for ambitious code changes and overnight experiments.
• Nathaniel Whittemore treats benchmarks as useful comparisons, not verdicts: scores help distinguish models when direct experience is limited, but one test or a combined index cannot establish how a model performs in every real use. Artificial Analysis has repeatedly changed its index as model capabilities and use cases shifted.
• Artificial Analysis v5 uses mostly private datasets to make benchmark gaming harder. It adds Frontier Code 2.0, judged by whether repository maintainers would merge the code, and Terminal Bench Science; it upgrades Humanity's Last Exam to HLE Diamond and refreshes AA Omniscience and CritPT.
• The new weighting reshuffled the leaderboard: Opus 5.5 scored 58, Astra rose from sixth to second with 57, and GPT-5.1 Sol climbed from ninth to third. Sonnet 5.5 fell from second to fourth, Terra rose from tenth to fifth, and Kimi K 3.1 Pro dropped from fourth to ninth.
• Frontier task costs vary sharply by model size and efficiency: GPT-6.1 Terra was cheapest, while Fable 5.1 was most expensive. Opus and Sonnet each cost about $10 per task; Astra cost $6.24, and Gemini 4 Carbon sat in the high-$7 range.
• Analysts disagree on whether V5's rankings should prioritize scientific research as heavily as agentic coding, and Nathaniel Whittemore says users may need category-specific comparisons. Critics also want leaderboards for medium, high, and extra-high effort settings, since max settings do not reflect typical use.
• Artificial Analysis added a Progress Index that tracks leading-model capability since November 2022. Lassan Al-Gaib used it to argue that OpenAI's strength comes from reinforcement learning or that its models perform poorly without reasoning; the same analysis estimated the US-China model gap at 3.3 months and narrowing.

Agents (2)

• V5 removes GDPVal AA in favor of AA Briefcase, AutomationBenchAA, and GDP.PDF, and drops Psycode and AA LCR. Its redesign reflects wider use of agentic coding and research, while improvements to compaction and memory have made long-context performance less decisive than it was in January.
• Artificial Analysis weights V5 evenly across work, coding, science, and general knowledge and hallucination performance. Coding tests now measure agentic work, and science tests measure agentic research, making the 25% science share a consequential choice rather than a neutral index setting.

Open Source (1)

• V5 widened the measured gap between leading US proprietary models and open-weight systems. Kyle Chan cautioned that Kimi's weakness is concentrated in science: it scored 34% on Terminal Bench Science versus Astra's 63%, yet scored 67% on Terminal Bench 4.0 and beat Claude on AutomationBenchAA.

Regulation (1)

• The DOJ's NVIDIA inquiry now spans more than its Hugging Face deal: Reuters sources describe scrutiny of venture investments, sales practices, customer pressure, and acquisitions including RunAI and SkedgeMD. The investigation also predates NVIDIA's Groq acqui-hire; no charges have been filed, and the probe could end without action.

Elections (9)

• The hosts said Trump canceled planned campaign stops in Kansas and Minnesota, then reversed course and scheduled rallies in Iowa and Kansas. Republican sources told the Washington Sun the rallies may be hurting candidates; polls showed Democrat Hamilton leading in Kansas, though the Washington Post warned its own poll had serious sampling problems.
• J.D. Vance answered a young voter's question about housing costs with a joke about Ohio politician John Husted, then touted a bipartisan bill restricting Wall Street home purchases that Trump had vetoed. The hosts said the exchange showed Vance ignoring the voter's concern and echoing Trump's campaign style.
• Vivek Ramaswamy’s campaign used an ad featuring Amy Acton’s estranged mother to attack Acton, then cut the ad from a minute to 30 seconds before removing it. The hosts said court records contradicted the mother’s claim that the family never lived in a tent and criticized the campaign for exploiting a visibly unwell woman and retraumatizing the family.
• Michigan Chronicle reporter Samuel Robinson said Abdul El-Sayed had momentum and expected overwhelming Black support in Detroit, while acknowledging some Republicans would vote for El-Sayed over his positions on Israel and Gaza. Robinson expected turnout to exceed the 2018 non-incumbent election and said early-vote returns could take until 2 or 3 a.m. to count.
• Robinson said Israel and Gaza are especially salient for Michigan Jewish and Muslim voters, but not the top issue for all voters. He described tension between some Black and Arab communities, warned that Arab American voters may be stereotyped through local disputes, and said Mike Rogers's campaign faced backlash over anti-Muslim rhetoric from some allies.
• Mai Vang and Angela Gonzalez-Torres challenged Democratic incumbents in California, with Vang saying AIPAC-aligned groups spent heavily against her over her Gaza ceasefire advocacy. Vang said Doris Matsui had loaned her campaign $7 million; Gonzalez-Torres said Jimmy Gomez had accepted substantial AIPAC support and used congressional funds for campaign mailers.
• Vang said Sacramento's district had shifted from a Democratic advantage of 22 points to 7 after Proposition 50, and described building a campaign with more than 1,000 volunteers. Gonzalez-Torres said DSA-LA helped organize canvassing and community campaigns; Vang cited a 700-person canvass organized with Sacramento DSA and UAW.
• California candidates said federal election probes and subpoenas were being used to cast doubt on voting, while a judge rejected an effort to block ballots from roughly 20,000 people alleged to be noncitizens. Vang said 22% of ballots had been returned in her district and urged eligible voters not to be intimidated.
• The hosts discussed two election tactics: a Texas Republican women's group sent students flyers urging gender transition to provoke trans panic, while Alaska's ballot-name controversy drew scrutiny over possible links to the Democratic Senatorial Campaign Committee and Mary Peltola's campaign. Emily said deceptive tactics can deepen public distrust in elections regardless of party.

Iran (1)

• Senator Chris Murphy called the administration's Iran war plans “insane,” warning that attacks on civilian infrastructure and Gulf oil capacity could trigger escalation and prolong damage even if the Strait of Hormuz reopened. The hosts also noted reports of missile defenses at Mar-a-Lago and military preparations for mass-casualty scenarios.

Custody (1)

• Binance disclosed that an attacker may have compromised a DevOps engineer’s laptop, gained access to withdrawal systems, and redirected a hot-wallet transaction through Ethereum’s EIP-7702 mechanism. The hosts reported that roughly $6.3 billion was stolen across blockchains in under 20 minutes, about $3 billion was frozen, and North Korean hackers were suspected.

AI Infrastructure (1)

• The hosts described a widening cyber risk: an AI agent reportedly accessed classified government systems through a contractor, while freezer systems were hacked at military commissaries and French grocery stores. They said six bases experienced freezer failures as the Pentagon prepared bases to operate without outside supplies for 90 days.

Markets (1)

• The hosts linked a broad tech-stock selloff to weak earnings and questioned whether AI investment had become a bubble. They cited U.S. third-quarter growth of 3.4%, with 12.7% growth in nonresidential fixed investment and nearly 0.9 percentage points of GDP growth from information-processing equipment, while the S&P 500 remained heavily concentrated in major technology firms.

Health (1)

• Vang said she would prioritize restoring Medicaid and food-assistance cuts, then pursue oversight of the administration; Gonzalez-Torres emphasized cash assistance, immigrant protections, and long-term cost-of-living relief. Both backed organizing for Medicare for All and getting corporate money out of politics, while Vang also named abolishing ICE as a voter demand.

Media (1)

• Ryan said John Kiriakou has blamed journalist Matthew Cole for helping expose him after Kiriakou disclosed CIA torture, though Ryan had no firsthand knowledge of that case. Ryan said Cole's handling of Reality Winner's document was a reporting mistake, but prosecutors had other evidence linking Winner to the leak; the episode reinforced the need to protect sources and limit document details.

BTC Markets (5)

• Jason Calacanis predicts Bitcoin will retest $57,000 within 12 months, citing ETF outflows and Bitcoin’s lack of cash flow; the host reads the call as an AI investor’s bearish challenge to Bitcoin.
• Willy Woo’s 10,000 Monte Carlo simulations put Bitcoin’s December 2028 price between $240,000 and $413,000 in the 28th-to-50th-percentile range, with scenarios spanning $116,000 to $1.4 million.
• Alex Thorn says Bitcoin has passed the point in prior halving cycles when bottoms formed, while the host says identical four-year cycles may no longer hold. Michael Nadeau and Thorn point to the 50-week moving average as a key early-bull-market level.
• Bitwise’s 2036 Bitcoin outlook ranges from $131,000 in its bear case to $3.6 million in its bull case, with a $1.6 million base case based on roughly 34% annual growth and Bitcoin taking a share of existing stores of value.
• Michael Saylor bought no Bitcoin during the week discussed after purchases fell from $140 million to roughly $20 million per week. The host also notes Bitcoin ETFs gained about 0.5% in assets over two weeks while Ether ETFs lost 6%, calling the divergence notable.

Fed (1)

• The FOMC held rates unchanged in a 10-to-2 vote, while September PCE inflation reached 3.6% year over year after three months of acceleration. The host links persistent inflation risk to $111 oil and possible renewed conflict with Iran after the midterms.

Macro (1)

• The host cites Michael Howell’s view that consumer confidence better reflects inflation pressure than official readings, with affordability a likely drag; a starter home was described as costing about $600,000 despite a recent one-month confidence rebound.

Markets (2)

• ICE and TradeXYZ plan a CFTC-regulated, U.S. venue for round-the-clock Hyperliquid perpetual futures, pending approval. Stanley Druckenmiller endorsed the venture, Elon Musk retweeted the endorsement, and Hyperliquid rose about 17% on the news.
• Robinhood reported $120 million in quarterly crypto revenue, including about $50 million attributed to Robinhood Chain, but chain revenue later fell to roughly $30,000 a day. Vlad Tenev said Robinhood would increase ecosystem and developer investment; meme-coin traders want activity that generates on-chain revenue.

Regulation (1)

• ICE and OKX also planned tokenized stocks for 24-hour trading, but more than 30 issuers sought removal under the SEC innovation exemption’s 30-day objection window. The host argues regulated, KYC-gated access could benefit issuers through more investors, liquidity, and trading venues.

Protocol (1)

• Zcash diverts 20% of block rewards to grants, split between an appointed committee and coin-holder votes. After Zcash’s price rise pushed the fund to $125 million, critics proposed ending or winding down the allocation because grants spend about $6 million yearly while roughly $68 million flows in.

Agents (1)

• Venice reported more than $1 million in VVV burns for October, funded by subscription fees, 5% of API-credit purchases, and discretionary monthly burns. The host sees buybacks as useful proof that a non-equity token has value, but warns that early companies may create more value by reinvesting revenue in growth.

Stablecoins (1)

• Ethena’s Earn Plus product inside Ethena Pay offers 12% yield on dollars; the host describes it as leveraged ENA rewards and suspects a leveraged basis trade, while noting the risk source needs closer scrutiny.

War (2)

• Luke Gromen says he underestimated how slowly Hormuz disruptions would affect the economy: China cut demand and drew down strategic reserves, while the United States and other Western countries released reserves. Fertilizer unaffordability could reduce crop yields next year, delaying food inflation while initially weakening crop prices.
• Gromen argues Iran can prolong the conflict by exhausting US air-defense missiles faster than the United States can replace them; he says Iran understands that resupply constraint and may receive help from Russia or China. He doubts the United States would deploy troops without air superiority.

Energy (3)

• Gromen says attacks on refineries could keep refining capacity constrained long after the conflict, even as reserve releases and China’s recovering purchases complicate the crude-supply picture. Diesel prices are nearing $200 a barrel in some regions, but he says he cannot explain why crude has not risen further.
• Ceresna said oil repeatedly recovered after headline-driven pullbacks, leaving crude in a decisive uptrend and potentially pricing in renewed post-election escalation. He identified $120 as a WTI upside reference if the trend continued; Gromen separately questioned how crude could stay elevated amid reserve releases and refinery outages.
• Ceresna noted that uranium equities were falling even as U3O8 held above $90, linking the disconnect to AI-related selling. He also said Brazil’s election surprised investors who expected Bolsonaro to win, triggering a sharp decline in the Brazil ETF before prices stabilized.

Iran (1)

• Gromen says Iran’s leverage comes from the bond market: rising yields could constrain US policy before the United States can force Iran to stop. He cited a 10-year yield that rose from 3.94% to 5.3%, then said it had eased.

Markets (4)

• Gromen called a 20-basis-point intraday fall in the 10-year yield unusually large and out of step with gold, oil, Bitcoin, and the dollar. He speculated that investors might anticipate election uncertainty or rapid fiscal cuts if the administration loses, but said he did not know the cause.
• Ceresna said gold lacked buyers near $4,000 and remained in a short-term distribution pattern, with a technical breakdown risking another 5% to 10% decline. He remained bullish long term but said a macro catalyst was needed for a trend reversal.
• Ceresna said crowded bullish positioning in copper had not produced a new high over three months, leaving the trade vulnerable if price broke below $650. He viewed a test near $670 as a decision point and stayed neutral until the chart showed accumulation.
• Ceresna said the USDA raised its corn production forecast, weakening the scarcity case after speculators had built bullish positions. He reported heavy selling through both long liquidation and new shorts, with corn prices falling as traders were shaken out.

Elections (1)

• Gromen says a contested or prolonged election could unsettle risk assets, while a Democratic sweep could leave spending largely intact through congressional bargaining. He sees both election uncertainty and AI-related credit concerns as possible reasons for the recent equity-market turbulence.

Labor (1)

• Gromen says AI-related job losses could weaken consumer borrowers’ ability to repay mortgages, credit-card balances, and auto loans. He cited layoff plans at Kroger, American Express, Barclays, and Microsoft as evidence that white-collar employment risk is spreading.

Banking (3)

• Gromen estimates that a 5% default rate on consumer debt would equal about $1 trillion, or roughly 40% of US banks’ $2.6 trillion equity base. He warns that AI-driven job losses could turn consumer credit into a larger banking risk than the AI companies’ own financing.
• Gromen says consumer-credit losses could force banks to restrict lending, issue equity, or sell Treasuries, adding pressure to government financing. He keeps 30% of his liquid net worth in cash and Treasury bills because he sees policy support for markets but cannot predict how the risks resolve.
• Patrick Ceresna recommended a Capital One bear put spread to express the risk that AI-related job losses could weaken consumer credit. He said the stock was near its 50-day moving average and that low implied volatility also made an outright put worth considering.

AI Infrastructure (1)

• Gromen says an AI investment bust could trigger recession, lower tax receipts, and sharply higher Treasury issuance; he argues a brief bond rally would likely reverse when borrowing needs become clear. In his scenario, the Fed would ultimately have to buy Treasuries, which he calls debt monetization despite official denials.

Models (1)

• Gromen says the administration has already restricted DeepSeek from government programs and their contractors, creating a protective moat for US AI firms despite its anti-regulation stance. He also says Washington discussed contingency plans for supporting AI if its financing collapsed.

Safety (1)

• Erik Townsend cited Anthropic’s IPO filing, which warns that advanced AI could pose catastrophic or existential risks without safeguards. Gromen connected that disclosure to the political conflict over whether the government should support AI as a national-security priority.

Society (1)

• Gromen warns that AI-driven job loss could fuel social unrest, especially if a growing, diverse US population faces fewer well-paid jobs. He expects the political consequences to intensify in 2027 and cites Peter Turchin’s account of elite overproduction and wealth inequality as recurring sources of instability.

Fed (1)

• Ceresna said the Fed delivered the hawkish hold markets expected, with no major repricing in yields. He said the dollar’s uptrend remained intact without a dovish pivot, while the euro near its 52-week lows and a break below 1.11 could drive further dollar strength.