1Tariffs and industrial policy are a voter-driven regime, not a policy phase
Michael Zezas, Morgan Stanley's deputy global head of research, spent an hour on The Compound arguing that the post-2020 policy regime does not revert. His mechanism is voter preference rather than personality: the best benefits of globalisation are behind us, on his reading, and the tariffs and industrial policy that followed reflect what electorates now want. If that is right, reshoring and reindustrialisation are a multi-year investable trend rather than a distortion to wait out, and geopolitical fragility becomes a standing input to markets rather than an occasional one. Deglobalisation came up seven times across the day's material.
The number attached to it is the part worth writing down. Zezas frames the AI build-out as the largest capex cycle "any of us will ever see", with Morgan Stanley estimating roughly $2.9trn of global data-centre spending across 2025 to 2028 — before anything is allowed for robotics or humanoids.
The week's single-name news is the same argument in miniature. ran 6–9% pre-market across several sessions on reporting of a Trump–Apple–Intel domestic-chip partnership, which is industrial policy showing up as a balance sheet. joined the $1trn club on memory and DRAM strength, rose with the storage rally, and jumped on a ByteDance deal. Reuters put equity inflows at a 19-month high, with record weekly inflows into technology funds; on Yahoo's markets wire, IPO anticipation around SpaceX was bidding up space ETFs and a Honeywell-backed Quantinuum listing was doing the same for quantum names.
Two things keep this a frame rather than a finding. It is one strategist on one podcast, and he named no tickers at all — the interview runs on macro and policy from end to end, with credit and municipal bonds referenced but nothing specific attached to them. And a multi-year thesis cannot be falsified by a single quarter, so the evidence for or against it arrives slowly, as capex lines inside earnings reports rather than as headlines about tariffs.
2Crude gave back 8% on a ceasefire, not on a deal
Brent fell roughly 8% on the week as the risk premium built during the Iran-Israel escalation came out of the price. The proximate cause was a truce rather than a settlement: an Israel-Hezbollah ceasefire in Lebanon, alongside talk of US-Iran negotiations, which Reuters and CNBC coverage read as risk-on for equities and bearish for oil.
What did not happen in the same week is the deal. The Swiss-hosted US-Iran talks were called off on Friday, and the wires that carried the de-escalation story carried "Iran deal doubts" next to it; the sanctions-relief politics remain contested enough that a French veto position at the UN was worth noting. The distinction matters because the two things move different variables. A ceasefire changes the probability that supply is interrupted. A sanctions settlement changes the quantity of supply. The 8% that came out of Brent is priced against the first, and it was the second that was called off.
Two further facts sit awkwardly beside a de-escalation read. The Wall Street Journal reported the Pentagon seeking about $80bn for the Iran war and associated bills, which is not the spending profile of a conflict being wound down. And the 60-day waiver on Hormuz transit fees is still running, so the chokepoint that carries the tail risk is operating under a temporary arrangement with a clock on it. On The Compound, the panel treated a full closure of the strait as low-probability while keeping it on the list of standing tail risks.
The consequence is an asymmetry in what it would take to reverse this. The premium came out on a truce holding, and a truce can stop holding in an afternoon; the barrels that would replace it need an agreement whose talks were called off on Friday.
3A $12.5m insider buy is the whole case for a $2bn biotech
The day's only conviction call on a single name is Ross Givens' pitch for (Aurinia Pharmaceuticals), and what carries it is insider behaviour rather than a product event. Kevin Tang, of Tang Capital, took control as unpaid chief executive, cut the board from nine seats to six, and owns about 10%. He bought $12.5m of stock in the open market on 2 June, and sold 10,000 January $15 puts — an obligation to buy roughly a million more shares if they trade down to that strike, which is accumulation by a second route.
The business underneath it: Lupkynis, the first FDA-approved oral treatment for lupus nephritis, is guided to $305–315m for the year and grew 25% in the first quarter, five years into its launch. First-quarter revenue was $77.5m with $34m of net income; there is around $379m of cash and a $15m buyback against a roughly $2bn market capitalisation, at a price-to-earnings ratio near seven. The pipeline adds an early-stage dual-pathway autoimmune programme and the assets from the Kezar Life Sciences acquisition. The chart argument is a seven-month base under about $17. Givens disclosed buying 1,000 shares himself.
The caveats belong to the call rather than to a footnote. This is a single-source retail pitch from a channel that spends much of its runtime promoting a paid subscription, and the idea went to paying members before it was published here — a reiterate to a wider audience, not a first airing. It is a single-name biotech, so the risk is two-sided and lumpy. And the recording is unreliable at the level of the company's own name: automatic captions render Aurinia as "Arena" and "Arbutus", and the ticker is confirmed only from what was shown on screen.
The case therefore rests on one insider's own capital rather than on a dated event, and nothing on the calendar forces a re-rating. Insider conviction is evidence about the insider; it becomes evidence about the business only if the revenue line keeps doing what it did in the first quarter.
What the sources recommended
Two videos produced one conviction call between them, both of the day's named calls sit on the buy side, and the second of them came off a news wire rather than out of anyone's research.
is Ross Givens' pick, set out above, complete with his own 1,000-share purchase and the paid-promotion caveat that travels with it. (Oracle) is on the list only because Zacks upgraded it to Buy, carried on Yahoo's markets wire — an analyst action rather than a conviction call from anyone who spoke on the day.
The reshoring argument gets a freight number on Tuesday and a memory number on Wednesday
US markets were closed on Friday for Juneteenth and reopen on Monday, which packs the week's tests into a short run of sessions. (FedEx) reports after Tuesday's close and is the freight bellwether — the closest thing on the calendar to a live reading of the goods economy the first section is about. (Micron) follows after Wednesday's close and is the cleanest read on the memory and AI-capex leg: a company that has just joined the $1trn club, reporting into a build-out estimated at $2.9trn. (Nike) reports the same evening, opens the week on Monday morning, and , and land on Thursday. (AeroVironment) reports the following Monday and is the defence-drone read on the rearmament theme.
The geopolitical tests have no times attached. The Swiss talks are stalled, the Lebanon ceasefire has to keep holding, and the Hormuz fee waiver keeps running down its 60 days. Any of the three turning would move crude faster than any of the earnings above.
