1The argument over AI and enterprise software has narrowed to whether per-seat licensing survives
The bear case on enterprise software is no longer that the demand disappears. It is that the unit being sold — a seat — stops making sense once the thing doing the work is an agent rather than an employee.
On Animal Spirits, Michael Batnick and Ben Carlson worked through a Kai Wu paper that sets today's software incumbents against earlier disrupted franchises: Blockbuster against , Borders and Radio Shack against Amazon. The uncomfortable finding is what happened to the fundamentals. Revenue per share in the disrupted names held up while the share prices fell, which means the damage arrived in the multiple rather than the income statement. A software incumbent can look fine on the numbers right up until the terminal value is repriced.
Felix & Friends put the counter-case within the same 24 hours, and it is not a demand argument either — it is a pricing-architecture argument. Felix's case for (ServiceNow) is that roughly half of new contracts are already written on consumption pricing rather than per seat, which decouples revenue from headcount, and that the workflow layer is what an agent needs in order to actually do anything — his phrase is that the brain needs a body. The supporting numbers are dated and checkable: Now Assist went from zero to $750m of annual recurring revenue in its first year and targets $1.5bn by year-end, against roughly twenty years for the core product to reach comparable scale, and a GSA contract puts the AI platform across federal agencies at a 70% discount. The stock is down about 56% year to date, at 60× trailing and 14.8× forward earnings.
The insider evidence is the load-bearing part of that pitch, and it is also where the caveats bite. CEO Bill McDermott bought $3m of stock personally and cancelled a scheduled selling plan to do it; four other executives bought; the company announced a $5bn buyback. Six members of Congress disclosed purchases — among them Ro Khanna, Michael McCaul and Josh Gottheimer, all sitting on committees touching cyber, AI or federal contracting — and the video notes that President Trump holds more than $1m of the stock himself. Against all of that: Felix does not own it, expects the downtrend may run sideways a while longer, and the episode is titled around a "$100 billion defense stock" that turns out not to be a defense name at all, with a paid course promoted throughout. The thesis is independently checkable and the framing is promotional; both are true at once. (Three of the five channels published nothing new into the holiday-shortened week, so this is two channels disagreeing, not a survey.)
One item from the tape sits inside the same argument. announced a $25bn debt-financed buyback and halved its free-cash-flow guidance in the same breath — an incumbent buying its own shares while lowering what it expects those shares to generate.
The question has therefore moved from whether AI takes software's demand to whether software can re-price fast enough to keep it. That makes the mix between seat-based and consumption contracts the number that settles it, not headline bookings.
2An index rule change can hand a loss-making listing 4% of the Nasdaq 100 within 15 days of its IPO
Batnick and Carlson spent a segment on SpaceX's freshly filed S-1, and the mechanical point is more interesting than the company. Nasdaq's fast-entry rule now admits a qualifying new listing to the Nasdaq 100 after 15 days, where the wait used to be three to six months, and with a 3× float cap applied they put the projected index weight at roughly 4–5%. Passive money would be committed on a calendar, before a single quarter as a public company.
What that weight would be buying: $18.7bn of revenue against a $4.9bn net loss. Around 200 companies in the S&P 500 carry more revenue than that, and has about five times as much. Starlink's average revenue per user has slid from $99 to $66 over three years as the customer mix moved from high-paying maritime and aviation accounts toward consumer and mobile-partnership subscribers.
One item cuts the other way, on the same morning. selected Starlink for its in-flight Wi-Fi, which is a win on the aviation side — the higher-paying end of the very mix the ARPU slide has been moving away from. A single contract does not reverse a three-year trend, but it lands on the side of the mix that pays.
The same rule applies to whatever lists next, and OpenAI and Anthropic are named as the queue behind it. A structural bid whose timing is knowable in advance and whose fundamentals are not is a new feature of owning the index.
3A Gulf escalation lifted oil and the dollar but sent gold to a two-month low
Reuters' Morning Bid led with the US-Iran ceasefire unravelling: further US strikes, IRGC retaliation, and shipping disrupted through Hormuz. The price response was lopsided. Oil rose and the dollar reached a one-week high, but gold fell 1.38% to $4,419, a two-month low — the opposite of what a fear bid does. Equity futures barely registered it, with S&P futures at -0.13%, Nasdaq futures -0.27% and Russell -0.29%, and the VIX up 2.33% to 16.67.
That combination reads as a currency and inflation event rather than a risk event. A stronger dollar mechanically pressures gold, and disrupted shipping is an inflation input before it is a growth shock. The rest of the tape corroborated the indifference: semiconductors led pre-market, with crossing a $1trn market capitalisation, up more than 5% on AI demand, higher on a ByteDance deal and upgraded to buy at Zacks — 12 of the 20 items in the pre-market news scan were semiconductor or AI stories. rose 13% on an earnings beat delivered despite the conflict hitting its sales.
Where the caution shows up is in positioning rather than price: CNBC reported bears building short exposure to small caps ahead of the coming economic data, and the Russell contract was the weakest of the three index futures. If the escalation is going to be repriced as a growth shock rather than an inflation one, gold turning back up is the first place it would appear.
4The flagged tail risk is 30-year sovereign yields, not a 9% equity drawdown
Animal Spirits opened on long-dated government yields going vertical across the UK, France, the US and Japan, borrowing the "fire alarm" framing from John Arnold's appearance on Patrick O'Shaughnessy's podcast. Batnick and Carlson both accept that the concern is real. Neither will say whether it cracks risk assets, on what horizon, or through which channel — and in the same episode they waved off the recent 9% equity pullback as "nothing". They do not agree with each other either: Carlson is the more cautious on valuation and historical comparison, and Batnick calls that framing lazy.
So the strongest-sounding claim of the day carries no number, no horizon and no position behind it. What would convert it is a long-end move that starts showing up in equity drawdowns rather than in podcast segments.
What the sources recommended
Two single-name buys came out of the two videos, and each arrived with a caveat from the person making it.
is Felix's pitch, covered above — down about 56% year to date at 14.8× forward earnings, with a GSA contract, a $750m first-year AI revenue ramp, and insider plus congressional buying as the confirming evidence. The caveats travel with it: he does not own the stock, he expects the downtrend may persist, and the episode is heavily promotional.
(Floor & Decor) is Michael Batnick's own disclosed purchase, and he presented it as deliberately contrarian rather than as a worked thesis — describing the chart as a "perfectly perfect downtrend", saying plainly that he does not like the stock, and buying it because it is oversold. The business is roughly 500 stores nationally. A personal contrarian nibble disclosed on air is a different object from a researched pitch, and he framed it as the former.
The per-seat question gets three answers after tonight's close
, and all report after the close — three seat-priced software businesses reporting straight into the argument in the first section, with estimates at $0.87, $2.90 and $0.08 respectively. reports alongside them at an estimated $3.00 and is the day's cleanest AI-server read-through. Before the open, ($1.58 estimated) is the consumer check, alongside a heavy China small- and mid-cap tape in (-$0.98) and ($22.53). Nothing on the macro calendar is time-stamped, which leaves Iran headlines as the live variable through the session.
