1The hawkish read survived a soft jobs print, in the rates market and in the gold tape
June nonfarm payrolls added only 57,000 jobs, taking the level to 158.98 million from 158.93 million. The unemployment rate fell anyway, to 4.2% from 4.3%. Those two facts pull opposite ways, and the rates market resolved them hawkishly: the 10-year Treasury backed up to 4.48% on 1 July from 4.44% on 30 June, and the 2s10s spread steepened to +0.35 from +0.30 a couple of sessions earlier. Fed funds sat unchanged at 3.63%, with a Federal Reserve press release landing across the 3rd to the 5th as the communication event underneath all of it.
Prediction markets say the same thing with more precision. Polymarket's July FOMC contract, the deepest of the three checked at roughly $38m of volume, prices no change at 89.5%, a hike at about 9.6% and a cut at 0.5% — so the residual risk into that meeting is a tightening, not an easing. The longer contract agrees: on about $40m of volume, no cuts at all in 2026 prices at 77.6%, one cut at 14.5% and two at 4.4%. Recession by year-end sits at 9.5%, but that sub-market turned over only about $2,400 in twenty-four hours against a $1.65m parent, so it is the thinnest number here and deserves the least weight.
The same hawkishness turns up somewhere less obvious. Clive Thompson, a retired wealth manager on Felix & Friends, puts gold's worst quarter in roughly thirteen years — the metal now at $4,187 after a 1.49% bounce — down to Kevin Warsh's appointment to the Fed and his reaffirmation of the 2% inflation target and the central bank's independence, on top of silver's January all-time high near $125 unwinding. His account of the mechanics is a stop-run: bullion banks pushed price through the stops clustered below roughly $120, then bought underneath the cascade. He has been adding to gold and silver modestly on the way down, and is openly skeptical of the dollar-reset thesis his own episode is titled around.
So the rates story and the metals story are one story: both are priced off whether that stance holds. A soft payroll number did not dent it, which leaves Fed communication rather than the labour market as the thing that moves either next — and the committee does not meet again until 29 July.
2The bear case on semiconductors is about the multiple, not the demand
Warren Pies, in clips carried on Excess Returns' weekly wrap, makes the sharpest version of the caution, and it is not a demand argument. Semiconductors are roughly 20% of the market and are being priced as if they were not cyclical: margins at and exploded, and price-to-sales expanded alongside them instead of contracting, which is the reverse of how peak-margin cyclicals have historically traded. His phrase is "the first real air-pocket of exuberance" — deliberately a bubble in one pocket of the market rather than a market-wide top. He dates the analogue as well: the 1990s run went 63 months and this one is about 30 months in, which is a way of saying he thinks it is early rather than finished.
The tape gave the argument something to chew on in both directions. Michael Burry's short of is the same peak-cycle thesis expressed as a position. fell as AI-demand worries collided with its guidance, was the most-active decliner at −5.25%, and slipped 1.39%. Against that, was bid on its addition to the S&P 500 and on AI-infrastructure demand, with and carried on the same theme. re-rated on a proprietary-data licensing deal with an AI lab — a reminder that the money in this build-out can accrue to whoever owns the training input rather than to whoever makes the silicon.
Meb Faber, in the same wrap, widens the frame: the broad US market is "very expensive" and can stay that way, and his answer is diversification — global equities, REITs, bonds, TIPS, commodities, equal-weight and non-mega-cap US. He treats bear markets as a feature that clears out speculation, and says your largest drawdown is always still ahead of you if you are doing this properly.
One caveat about how both of them reach a listener: Pies and Faber arrive through the same weekly wrap, which is one programme's clip selection rather than two independent bookings. The agreement between them is a curated pairing, not corroboration. The consequence of their shared point stands either way — a strong demand print does not answer the objection, because the objection is about price.
3The bigger risk to the AI build-out is political, not technical
Pies' second point is the more unusual one. The risk to the AI trade that gets the airtime — open-source models and token-efficiency gains undercutting the economics — is, on his read, the wrong one to watch. The one that matters is political: organised backlash against data centres, already visible on both the left and the right, with 2027–28 and the midterms as the window it opens in.
The mechanism runs through financing rather than through regulation directly. If public sentiment sours, lab funding gets harder, model progress slows, and the pace of that progress is what the bull case actually rests on. That makes it a slower risk than a valuation de-rating and a harder one to hedge, because it carries no print date. It also surfaces first in permitting decisions and local politics, which is not where a market watches for it.
4Cheap once-yearly injections are the live threat to legacy pharma's margins
Andrew Dudum, chief executive of Hims & Hers, told Ed Elson on Prof G Markets that the company sells branded GLP-1s at about $150 against roughly $1,500 through the brand channel, on the back of owning its own provider group, its own records system and its own compounding facilities. Compounded GLP-1s under the 2024 shortage exemption "doubled the company overnight", and revenue has gone from roughly $100–200m to a guide near $3bn this year. now partners directly with and , the makers it was previously working around, with side-effect titration as the wedge that got it there. Take that as a founder's account of his own company, which is what it is — an interview with no bear case put to the guest.
The industry-level claim is the more checkable one. has an early-trial once-yearly cardiovascular injection at something like $30–50 a shot, aimed at the same LDL problem that daily statins and 's Repatha treat. If a once-a-year injection at that price displaces a chronic-dosing product, what is lost is not share between two firms — it is the price of a category. Dudum's framing is that "the age of unrestrained pharma margins is gone", with government-negotiated pricing as the mechanism, and he notes generics already arriving in Canada with US semaglutide roughly four years out. He also puts GLP-1s ahead of AI for societal impact over the next five years, on the downstream effects in diabetes, cancer, cardiovascular disease and addiction.
The near catalyst is regulatory. A July meeting is expected to weigh reclassifying peptides including BPC-157 and TB500 from category 2 back to category 1, which would open consumer access, and the legacy industry opposes it. That is a datable test of the distribution argument above, and it lands this month.
The next Fed test is three weeks out, so this week's are smaller ones
US markets reopen on Monday 6 July after the Independence Day weekend, with Friday 3 July observed as the holiday, so the gap has to price the hawkish Fed read against continuing US–Iran de-escalation — which has taken oil back to pre-war levels — in a single print. One force tightens financial conditions and the other loosens them, and the reopen makes them net out in one number.
The decision itself is not until Wednesday 29 July, currently priced at 89.5% for a hold, which leaves three weeks in which communication does all the work. Earnings give little in the interim: the 6th and 7th carry closed-end funds only, then (Levi Strauss), (ConAgra), (Helen of Troy), (PriceSmart), and (Kura Sushi) report on the 8th, with (Delta) on the 10th opening the season unofficially.
