Why the Midterms Matter for Market Liquidity | Macro Mondays: Nov. 2, 2026

Anthropic

Across the shows: 81 speakers in other episodes, 96 bullish and 63 bearish calls, past 30 days

• Andreas Steno · Steno ResearchBullish

  Anthropic's growth is the fastest in history; market's view that its margins go to zero is wrong.

  Andreas cites Anthropic's climb from about $1B to about $75B in annualized recurring revenue, which he says is the fastest ever for a company even after adjusting for inflation. He argues the market is pessimistic about the compute commitments behind that build-out and dismisses the consensus that Anthropic's product margin will go to zero, which he says he hears in every client meeting. He concedes the Q3 numbers showed some lost momentum versus H1 but still look good overall. He admits some of the pessimism may have elements of truth.

  “I find it interesting that the company that without doubt has moved from 1 to 75,000,000,000, the fastest ever in history, even when you adjust for the inflation of money.”at 16:03

• Andreas Steno · Steno ResearchBearish

  Anthropic's compute scarcity is hurting its growth trajectory, so he prefers OpenAI.

  Andreas says he is reluctant to buy Anthropic's IPO because of its second derivative of growth. Anthropic is compute constrained, and he notes that spot GPU and data center prices have surged since October, which raises Anthropic's costs and limits capacity. He has said since June that he prefers exposure to OpenAI over Anthropic for this reason. He is explicit that this is not a dislike of Anthropic, which he calls the best business ever.

  “I prefer to be involved in OpenAI than Anthropic, basically due to the second derivative of growth.”at 20:43

OpenAI

Across the shows: 28 speakers in other episodes, 13 bullish and 21 bearish calls, past 30 days

• Andreas Steno · Steno ResearchBullish

  OpenAI is the preferred AI exposure over Anthropic because its growth trajectory is stronger.

  Andreas says he likes being involved in the AI sector but prefers OpenAI to Anthropic. His reason is the second derivative of growth, with Anthropic facing compute constraints that the spot GPU price surge makes worse. The preference is relative: he prefers OpenAI's growth profile, and he does not describe a view on OpenAI's absolute prospects.

  “I prefer to be involved in OpenAI than Anthropic, basically due to the second derivative of growth.”at 20:43

US elections & politics

• Mikkel Rosenvold · Steno Research (Real Vision)

  Blue wave looks too good to be true; Republicans have a real chance to hold the Senate.

  Mikkel says the Senate is where Republicans have a decent or some chance of holding their majority, and Michigan is a state they need. He has seen Rogers gain some momentum, though it is turning around. He is skeptical of the supposed blue wave for Democrats, which seems too good to be true. He also argues a Democratic sweep would have little market impact because it is what everyone expects.

  “I'm a little bit skeptical about the blue wave that we're supposedly seeing. It seems a little too good to be true for the democratic side.”at 10:40

• Andreas Steno · Steno Research

  Republicans can only surprise positively versus prediction-market pricing of Democratic midterm gains.

  Andreas notes prediction markets expect the average contested seat to move about 12.7 percentage points toward Democrats versus 2022 or 2024, which he calls incredible and fully priced. He sees the surprise potential as a Republican rebound in the Senate. He cites Michigan, where Haley Stevens withdrew support for Elissa Slotkin's opponent, with a brief market move that has not lasted, and Ken Paxton receiving huge funding from Elon Musk in Texas. Given extreme positioning, he thinks a surprise would wrong-foot markets.

  “In short, I'm of the view that the Republicans can only surprise positively versus this pricing.”at 12:46

AI infrastructure & data centers

• Andreas Steno · Steno Research

  Market's uniform pessimism about AI capex is overdone, unlike the lead-up to 2000.

  Andreas reports that credit and equity investors at a large conference, and clients generally, say no more capex can be funded for AI. He argues that the sector delivering the fastest growth in history is getting the opposite of the response seen before 2000, when investors were eager to fund build-out. He says he is not ruling out that some of the concern has merit, but he finds the uniform pessimism around the best case ever seen incredibly wrong.

  “I just find it incredibly interesting that we see such uniform pessimism around the best case that mankind has ever delivered.”at 19:19

• Andreas Steno · Steno Research

  Spot prices for GPUs and data center capacity are rising sharply as capital for new compute dries up.

  Andreas says GPU prices have been on an absolute tear since the start of October. Investors are no longer willing to fund new compute, so the risk of extra capacity coming online has fallen, and existing compute is repricing upward. He points to H200 spot prices as a straight-line rise. He calls the rise across GPUs and broad data center capacity a fair assessment of the market.

  “So the spot price of everything from GPUs to broad data center capacity is up a lot.”at 21:49

Middle East & Iran

• Mikkel Rosenvold · Steno Research (Real Vision)

  Iran is ready for peace with bigger concessions than in summer; Trump earns some credit.

  Mikkel cites gradual steps over recent weeks as Iran eases its stance and becomes more willing to negotiate. He argues Iran is legitimately afraid Trump will launch a big attack after the midterms, which he says is almost planned, so it is trying to get a deal in beforehand. He thinks a deal probably won't be done in time for the midterms, but he says the bottom line is Iran is ready for peace, with a deal containing more Iranian concessions than the summer. He credits Trump for finding a way out of a hole, at a huge cost to markets.

  “Iran is ready for a peace. They're gonna have much lighter peace deal or a much... A peace deal that contains more concessions from the side than what we have in the summer.”at 3:12

Market sentiment, positioning & volatility

• Andreas Steno · Steno Research

  October's deleveraging is the biggest in a decade and may be near its final hours.

  Andreas says Morgan Stanley prime brokerage data shows the average hedge fund took off notional exposure in October, the biggest deleveraging in about ten years. Positioning is extreme: the market is incredibly long oil, incredibly short bonds, net short everything Nasdaq-related, and net long dollars. He points to a hot PPI print after which yields fell about 35 basis points, a move against the data, as a turning point that triggers degrossing. He says we are approaching the final hours, though markets then take time to rebuild confidence.

  “it's probably the biggest deleveraging event that we've seen basically over the past ten years.”at 5:38

Oil & fuels

• Andreas Steno · Steno Research

  Falling oil on Iran talks is unwinding a crowded long and could speed the deleveraging's end.

  Andreas notes oil is dropping sharply after the Iran leadership green-lit negotiations. He says the market is incredibly long oil as part of its crowded positioning. He argues the deleveraging could be fast-tracked if the oil price keeps sliding as it did that day, and that sequential progress toward a peace deal should alleviate the month's concerns.

  “It could be fast tracked if the oil price continues to slide the way that it's done today.”at 8:29

US tax & spending policy

• Mikkel Rosenvold · Steno Research (Real Vision)

  The lame duck period could be more productive than the past two years, with appropriations and debt ceiling action.

  Mikkel notes Trump has passed few major bills beyond the big beautiful bill and a few others, with the rest being appropriations and minor naming bills. He expects the administration to push hard in the lame duck to pass 2027 appropriations, including defense appropriations, and possibly raise the debt ceiling now. That would remove a big obstacle or risk for markets. He says it will be tricky to win over senators like Rand Paul, but Trump has an incentive to act before Democrats take over, and a Democratic sweep would make the next two years very tough.

  “I have a feeling that the upcoming lame duck period... could be even more productive than the past two years have been since Donald Trump took off.”at 13:43

AI bubble & tech valuations

• Andreas Steno · Steno Research

  Tech is priced below its 2022 bottom on forward earnings, so the cycle can run longer.

  Andreas notes that the tech sector is currently priced on forward earnings lower than it was at the 2022 bottom. He reads the market's conservatism about paying for future earnings as prudent. In his view, a glass-half-empty market allows the cycle to run for longer than an euphoric one would, and he contrasts this with a bubble-like scenario.

  “We're currently pricing the tech sector in forward terms lower than we did during the bottom in 2022.”at 18:00

US–China AI & chip race

• Andreas Steno · Steno Research

  China cannot sustainably deliver free AI compute; zero-margin pricing would trigger Western bans.

  Andreas says the notion that China can deliver compute capacity for free to everyone is wrong. He acknowledges Chinese providers may run zero or negative margins to win market share. He argues that if that happened, everything China-related would likely be banned from entering the West in an AI context. The argument is that this scenario is a risk to Chinese players, not a durable pricing reality.

  “this notion that China can just deliver compute capacity for free to everyone is utter bullshit.”at 23:12

AI pricing & unit economics

• Andreas Steno · Steno Research

  LLM prices must cover compute costs plus a margin, so sub-cost pricing makes no sense long term.

  Andreas accepts the argument that LLMs will be commoditized with no moats, but argues providers must still pass on at least their compute costs. He notes Anthropic has passed on input cost increases several times over recent quarters. He rejects the notion that LLMs will be priced below compute cost over time, since no one would do that sustainably. He says current pricing is a compute cost plus a decent margin.

  “Why on earth should anyone be willing to do that? I mean, it just doesn't make any sense over time.”at 23:59

AI chips & semiconductors

• Andreas Steno · Steno Research

  Semiconductors look great fundamentally, and his fund is positioned for the next leg up.

  Andreas says the firm still thinks the semiconductor space looks great fundamentally. He sees the next upward leg depending on a slightly more benign inflation picture and bonds behaving better, which he says is starting to show but is early days. The growth rate of change must also improve, and he points to the ISM and a US growth nowcast showing signs of life. He thinks semis could again become the consensus only play within a month.

  “In all honesty, we still think the semiconductor space is looking great.”at 26:31

US growth & recession risk

• Andreas Steno · Steno Research

  A benign ISM beat and improving nowcast suggest US growth momentum is turning up.

  Andreas says ISM manufacturing printed higher than expected, with activity and orders up, prices down, and employment still solid, which he calls a benign report across the board. He views the ISM as a great first step toward an improvement in the rate of change in growth, which the market needs for the next upward leg. He also says the firm's nowcast on US growth is showing signs of life.

  “The ISM manufacturing was a great first step in that direction today. Our nowcasting on US growth is showing signs of life.”at 26:42