1Software's premium to the S&P 500 flipped to a discount, and the argument moved from price to moats
Software stocks now trade at roughly a 10% discount to the S&P 500 — the first time on record, against a premium that has averaged about 32% over the past two decades. The software ETF is down around 30% peak to trough. That is the headline number, and Kai Wu of Sparkline Capital, speaking on Excess Returns, spent an hour arguing it is the least interesting thing about the cohort.
His case is that the dispersion inside the discount does the real work. and are down somewhere between 50% and 80% from their highs, and sits in the same badly drawn-down basket. Cheap software bought because AI fear made it cheap is, on his framework, a value trap: the survivors will be identified by intangible assets rather than by price — customer loyalty, brand equity, human capital, network effects, switching costs. The line he keeps returning to is that the code is not the moat.
The evidence base is unusual for a podcast. He builds on David Teece's "complementary assets" work and a 200-year dataset of technological disruption, and his historical cases are companies that survived without owning the best technology — Coca-Cola, IBM after the PC, GE. In each, the ecosystem and the network did the surviving, not the invention.
One item in the day's news is that thesis being priced live. Indian IT services stocks (, ) had their worst session in four months on AI-disruption fears, per Yahoo Finance — a services cohort with real scale marked down on the argument that the work itself is automatable.
Note what this is not. Wu named no positions in either direction; the framework identifies a question, and the question is which software companies get disintermediated and which are protected by something other than their code. A cohort at a first-ever discount is that question priced before it has an answer.
2The AI hardware trade stopped moving as one block
Tuesday split the AI hardware complex rather than lifting it. rose about 19% on server demand, and about 20% after an endorsement from Nvidia's Huang — both per Yahoo Finance and Google News. In the same session fell 9% and fell 4%. was up 7.84% in Wednesday's pre-market. Six of the fifteen most recent Google News items were semiconductor names, and they were not pointing the same way.
The rest of the tape was quiet around it. US futures were narrowly mixed pre-market — the S&P down 0.03%, the Dow down 0.24%, the Nasdaq up 0.08%, the Russell down 0.31% — with the VIX at 16.12, up 2.2%, and gold at $4,484, off 0.8% from recent highs. Tokyo's Nikkei 225 closed at an all-time high overnight.
Two more markers of the same dispersion. crossed into the $1 trillion market-cap club. And part of the bid rotated offshore: rose about 6% on an AI model launch, and led a broader Chinese technology rally, per Google News.
One driver is contested. Yahoo Finance recorded a Qualcomm–ByteDance partnership as a lift on the same day Google News had down 9%. Both cannot be that name's story for the session; the price is the part that settled.
and its GTC event were still carrying the headline narrative from Monday. But a theme in which one name gains a fifth of its value while another loses a tenth on the same day is no longer one trade. Dispersion that wide means basket exposure and single-name exposure inside AI hardware have stopped being the same bet.
3Copper's move to $6.67 a pound is being called the start of a super-cycle, on chart evidence alone
Ross Givens, on his own channel, argues that physical copper's run from $4.50 to $6.67 a pound over the past year is the first leg of a multi-year super-cycle, and puts $10–15 on it over two to three years. His analogue is 2002–06, when copper went from about 60 cents to $4, and his reading of the weekly log-scale chart is that it looks "almost identical to gold two years ago."
He holds the view himself in futures — three contracts — but pitches miners as the retail expression, on operational leverage: by his arithmetic a 50% move in the metal roughly doubles miner profit. The three names come with levels attached. breaking out of a $195–200 base, stop at $187 for about 6% of risk, target $250–300. around $70 with an 8–10% stop and a $90–95 target. as the fast one, already a few percent past his ideal $29 entry at $31, with a stop he himself describes as an arbitrary 10%.
Two things travel with the call. There is no fundamental work behind it beyond that operational-leverage arithmetic — it is a chart-pattern and cycle-analogy case. And the calls are interleaved with a repeated pitch for a $5-a-year paid course, in a video titled "These 2 Stocks Are About To Go PARABOLIC" that in fact names three.
One catalyst listed for is an Iran-driven accumulation pattern, which sits awkwardly against the same morning's news, where Yahoo Finance had Iran tension easing and the pre-market buoyant partly because of it. Those two do not point the same way. What settles the thesis is the metal itself: at $6.67 the operational leverage that makes miners attractive on the way up works identically on the way down.
4The IPO window reopened at the speculative end of the market
Quantinuum, the Honeywell-backed quantum computing company, debuted, and Yahoo Finance framed the listing as a test of the sector rather than a coronation. Alongside it, an IPO frenzy around SpaceX pulled space stocks and related ETFs up with it. Two listings, one window.
Felix, on Felix & Friends, was pitching into that window the same day. is his quantum name — "trying to be the Intel of quantum," in his framing — and (Beta Technologies) his electric-aircraft one. The Beta numbers are the useful part, because they show what the window is pricing: UPS and UTI committed to 991 aircraft, a GE Aerospace partnership on a hybrid-electric turbogenerator, and 123 charging stations built, against $10m of quarterly revenue and a $39–40m full-year guide at roughly a $4bn market capitalisation. That is about a hundred times revenue for a business still converting orders into deliveries.
He calls it a high-risk, high-potential bet, and the execution risk is acknowledged rather than worked through — no valuation case is shown for any of the six names on his list. The video also runs a repeated pitch for a free live session that feeds a paid product, under the title "Leaked: Last EASY Wealth Opportunity in 2026." That does not make the names wrong, but the framing is selling something. What marks this window is where the debut trades once it settles, not where the pitches point.
What the sources recommended
Fifteen names sit on the buy side, and only nine came from a person making a call — the other six were put in the column by their own price action in the day's news.
The stock pickers. Ross Givens supplied the three copper miners above: , and . Felix supplied six. , which he calls "the most underestimated AI story," on the argument that Oracle's database and cloud layer sits underneath the AI applications rather than competing with them. (Dynatrace) as the observability beneficiary of enterprises building AI systems. (Tenable) in cybersecurity, on a vulnerability-management angle rather than perimeter defence, and the one name of the six he says he bought with his own money. (10X Genomics), his "secret favourite," in single-cell genomics, an industry he calls bigger than AI. Plus and from the section above.
is the only name with a second, independent source behind it: Zacks upgraded it to buy the same day.
The momentum entries. , , , , and come from the day's news flow rather than from anyone's thesis. They are in the buy column because they moved, which is a different kind of evidence, and a shorter-lived one.
The day's strongest argument stopped short of positions. Kai Wu's value-trap examples — GoDaddy, Wix-style domain registrars — are illustrations of a category, not calls, and reading them as sell recommendations would be reading a framework as a trade.
The software and hardware reads both get tested after tonight's close
reports after the close, and the number that matters for the first section is Agentforce — the AI product meant to answer the disintermediation question, at one of the two names the cohort discount is built on. reports the same evening and is the bigger tape-mover: a mega-cap semiconductor print into a complex that just dispersed. , , and C3 AI follow after the close, and reports before the open.
Tomorrow morning brings (Ciena), the cleanest optical-networking read on AI infrastructure spending, then DocuSign, Samsara and Rubrik after the close. Somewhere inside the next seven days sits what Yahoo Finance called the most important data release of the month, which its summary did not name.
