1The AI build is being paid for out of cash flow, and that is the number the 2000 comparison fails on
Denise Chisum, Director of Quantitative Market Strategy at Fidelity Investments, on The Compound's TCAF, moved the bubble question off valuation and onto the balance sheet. Aggregate corporate capex is running at less than one times free cash flow. At the 2000 peak it was three-and-a-half to four times. Companies are spending inside what the business generates, which is the difference between a capital cycle and a credit event — and unlike an argument about multiples, it is a number that can be re-checked every quarter.
The spending is also broadening rather than concentrating. On her count all eleven GICS sectors are accelerating capex-to-sales, not just the hyperscalers. Goldman Sachs raised its S&P 500 target to 8,000 from 7,600 on the same logic — AI-driven earnings growth, FY26 EPS growth around 24%, and a market it reads as sober rather than bubbly. There is evidence the spend is converting, too: Meta's Reels is now running at roughly $50bn annualised, larger than Netflix, which is capex showing up as consumer monetisation rather than as a promise. Brown, on the same show, put it less delicately: AI, he said, "literally rewired our brains."
The awkward fact she does not dodge is the semiconductors. The complex sits about 70% above its 200-day moving average, which does not appear anywhere in 25 years of her data; the single-name version of that extension is , which joined the trillion-dollar club as UBS lifted its price target to $1,625 from $535. Her answer is that earnings have compounded faster than price over the trailing two years, so the extension is earnings-led rather than multiple-led, and that from setups like this one the following twelve months have historically outperformed 65% of the time. Note exactly what that is: a base rate with a one-in-three failure branch, not a forecast. The related trade she says has already been collected — tech's forward P/E converged with the broad S&P earlier this year, and the 38 days since delivered 16% relative outperformance.
If a setup like this breaks, the conventional route is the Fed, and the week's heaviest news flow pointed that way: the US–Iran ceasefire framework ran across roughly fifteen items, mostly Reuters-sourced, with a Trump decision flagged for Friday and a demand that Hormuz reopen. Brent posted a monthly drop on the truce extension while the S&P 500 and Nasdaq closed at record highs. Chisum's argument is that oil is the wrong route: an oil shock has passed through to core inflation less than half the time since the 1980s, so it arrives as a tax on the consumer rather than a trigger for a hike, and the equity market's historical sweet spot is 3–4% inflation. The counterweight is live and domestic — April PCE recorded its largest annual increase in three years. The oil path to a hike is weak on the record; the home-grown path is the one still open.
2The year's index gain has come from outside the mega-caps, and the small-cap earnings recession has begun to inflect
The Magnificent Seven are up 6.6% year to date against 10.25% for everything else, and the top ten stocks contributed only 2.3 percentage points of the index's 10% gain. On Chisum's reading that divergence is health, not decay: the market is rising on the other 493 rather than on a narrowing handful.
What would make that durable is earnings, and this is where her strongest numbers sit. Equal-weight earnings are inflecting after what she calls a three-year "full-employment recession" — a profits downturn that never produced a labour-market one — leaving the possibility that we are four months into a three-year recovery rather than late in an expansion. Russell 2000 EPS is forecast up 30%. Small-cap financials are expected to grow earnings 66% against 7% for their large-cap equivalents. ISM new orders have inflected for the first time in three years. Margins outside tech are expanding on unit-labour-cost discipline and productivity, and the single best correlate of forward margins in her work is the unit-labour-cost quartile, currently the bottom one.
This is one strategist's framework, delivered at length on a podcast rather than tested against a counterparty: the inputs are checkable, the reading of them is hers. The claim also carries its own scoreboard: if the index gain is genuinely coming from the other 493, it has to keep showing up in their prints, and small-cap earnings are where it either lands or doesn't.
3The AI power bottleneck is a 2030s build with 2026 tickers attached
Felix, on Felix & Friends, set out the constraint plainly: AI is the most power-hungry technology yet built, the US grid is "buckling", data-centre demand is forecast to quadruple by 2030, and the only round-the-clock baseload answer he sees is nuclear. He mapped that into a five-layer chain and named a stock at each layer — for fuel and logistics, for reactor design, for the regulatory moat, for power conversion, for data-centre cooling and distribution.
The dates buried inside those picks are the story. 's first test reactor is targeted for July 2026, with commercial operation, in his own words, "ages away"; it holds $2.5bn of cash and has selection for a Department of Energy surplus-plutonium programme plus partnerships with Meta, Nvidia and Equinix. owns the only NRC design-certified small reactor, and its deployments are 2032–2033. holds $550m of cash against zero revenue, and its qualifying event this cycle was a $13m acquisition of a transport company — buying "the railway for the industry", as he framed it. The two names in the chain that already earn money are the two selling into the build happening now rather than the one scheduled for the next decade: , which makes the power-conversion modules sitting between the wall and the GPU, has raised guidance and signed a royalty deal, and , framed as "the Cisco of the AI buildout", with Microsoft, Amazon, Google and Meta as customers.
He attached his own brakes, which is more than the format usually gets: a 2–5% maximum position size, a written exit rule before entry, and an explicit statement that this is the base phase of a climb-top-collapse pattern. He twice called the pre-revenue names — , and — high-risk plays not to bet the farm on. The video also functions as a funnel: a paid workshop is pitched more than five times through it, alongside a lottery-style teaser about politicians buying the same stock. The constraint he describes is dated to the 2030s and the securities offered against it are priced today; that gap is the risk, and it is why he wrote exit rules onto his own picks.
4A $1.75 trillion listing is two weeks out, and the biggest claim attached to it rests on a prediction rather than an approval
Ross Givens has SpaceX listing on 12 June under at roughly a $1.75trn valuation — the largest IPO on record, oversubscribed, and expected by him to gap 30–50% on day one. His own title calls the IPO a trap, and the distinction is deliberate: the listing is not where he thinks the money is.
The money, on his account, is in what follows. He carries a prediction from Chamath Palihapitiya on the All-In podcast that SpaceX rolls into Tesla by end-2026 or mid-2027, producing an "Elon Corp" of roughly $3.4–4trn that would be about 5% of the S&P 500 and 7.5–8% of the Nasdaq-100. Against $19trn of passive assets, he puts the forced buying at index inclusion at $500–700bn. That is the engine under at high conviction, under as pre-IPO exposure, and under , a UK-listed small-cap radio-component supplier.
Three things sit awkwardly with it. The merger arc traces to one podcast prediction, and the private-company board approvals it would need are not in evidence. The index-inclusion arithmetic assumes both the timeline and the ratio rather than deriving them from a published rebalance schedule. And the flow argument cuts both ways, which his own framing acknowledges: a weight that large has to be funded, so the same rebalance that bids for the new entrant mechanically pressures the mega-cap weights it dilutes — a near-term drag on the complex before it is a bid for one member of it. The segment also carries a $5 subscription service pitched twice.
Set that against the week's other argument about passive money, which points the opposite way. Robert Hagstrom, author of The Warren Buffett Way, on Excess Returns with Bogumil Baranowski, argues the market's machinery has outgrown the businesses beneath it: options notional value now exceeds the entire US equity market capitalisation on a daily basis, and there are more indices and ETFs than there are listed stocks. His frame is Buffett's cathedral and the casino — the business-driven investor visits the exchange to transact, then returns to the analysis — and his own Global Leaders Portfolio is the deliberate counter-example, 21 stocks on a 4.4% owner-earnings yield, 32% ROIC and 17% forward sales growth. Recorded in the same conversation without a call attached: has been flat over twelve months while the market gained double digits, through the Greg Abel transition. So one channel treats $19trn of passive money as the engine of a trade and the other treats that same scale as the reason prices have detached from ownership. Both cannot be the right way to use the fact.
The other space story of the week was a failure. Blue Origin lost a rocket on 29 May, read as a setback to 's Project Kuiper satellite-internet ambitions and, by the same token, a tailwind for second-tier satellite communications on Defense Department diversification — and — with up 12% on the spillover.
What the sources recommended
Nine single-name buy calls came out of four videos, and both of the channels that made them ran paid promotions inside the same videos — a workshop funnel pitched more than five times in one, a $5 subscription service twice in the other.
Five are the power chain from Felix & Friends: and at high conviction, , and at medium, all on a months horizon, and all subject to the 2–5% sizing cap and pre-written exit rule he put on them himself. Three of the five — , , — are pre-revenue, and he said so.
Four are the SpaceX complex from Ross Givens: on the 12 June listing, on the merger arc, as a pre-IPO backdoor, and on rocket components. Two of those carry performance claims that are the channel's own account of its earlier alerts rather than anything independently recorded — "nearly tripled in 48 days" previously, "doubled in under a week".
All nine sit with the promoted channels, and that concentration is the most useful thing to know about the day's calls. The two sources with nothing to sell dealt in frameworks and aggregates instead: Excess Returns spent its hour on market structure, and The Compound's numbers were index-level and sector-level throughout.
The broadening thesis gets its first prints on Monday
reports Monday on AI optical, which is the cleanest single read on whether the capex flow in the first section is still accelerating at the component layer. and land the same day, and 's distribution volumes are the kind of thing the manufacturing-recovery claim either shows up in or doesn't. Tuesday brings , plus and — two consumer prints, which is where a tax on the consumer would surface before it reached anything else.
The Iran framework decision was flagged for Friday with weekend follow-through still pending, and the record closes that preceded it came on ceasefire optimism — so the constructive outcome is already partly in the price. And the listing the fourth section turns on has a date rather than a thesis attached — 12 June, two weeks out, which makes it the rare claim with a scheduled mark.
