Hi all,
A quick note before this week’s reads. As I mentioned on Monday, Giles Capital is going paid. This edition, the midweek Reads, stays free and will keep landing every week, so this email won’t change. It will only get better as I spend more time on it.
From this coming Monday, 24 August, the Giles Capital Weekly update moves to paid: every write-up, the summaries, the fund letters and increasingly my own analysis on the best ideas. If you’d like to keep receiving that, there’s an early rate of $20 a month (or $200 a year, about $16.60 a month) available until 30 August, after which it’s $25 a month or $250 a year.
Now, on to this week’s reads..
Capital allocation is the thread. Featured this week: the AI buildout economics, Japan through two different lenses, private credit stress showing its first fractures, and company analysis from Korea to the UK to Singapore. ‘The One to Save’ this week is Ben Thompson and Patrick O'Shaughnessy on who wins as the technology spending cycle meets its first revenue test.
Ben Thompson argues Nvidia's Risky Business is structural: the cloud giants spending most on AI chips are also working hardest to replace them. The longer version of this thesis is this week's One to save. (10 min read)
Eagle Point Capital on reinvestment returns by sector: the capital expenditure that compounds versus the kind that subsidises competition, with examples across technology, energy, and industrials. (8 min read)
Nick Maggiulli tests staged versus lump-sum selling for a concentrated stock holding, with tax sensitivity analysis that is genuinely decision-useful for anyone managing a legacy position. (9 min read)
The Nikkei 225 tilts toward exporters and misleads on domestic Japan, while the Yomiuri 333 weights local consumer and services names and has quietly outperformed. (4 min read)
A ProPublica investigation into how a $533 million artillery plant was funded by taxpayers, failed to produce a single shell, and generated zero consequences for the contractor. (15 min read)
Peter Attia on multifactorial trials: why bundling multiple interventions into one study makes results unattributable, a point that transfers directly to evaluating any manager who changes several things at once. (7 min read)
Vishal Khandelwal on why optionality as a trap is the more useful frame for decision-making, and the case for self-imposed constraints in investment and in life. (7 min read)
The S&P500's 16.9% net margin is real but concentrated: this piece examines the margin by sector and shows how much of the expansion lives in a handful of technology names. (7 min read)
Bank of England researchers model AI earnings spillovers: what a material miss in big-tech results would mean for global equities, bonds, and credit spreads. (9 min read)
CoreWeave's lenders had a structurally better quarter than its equity holders: alphaseeker84 explains CoreWeave's capital structure and where the margin risk concentrates at each layer. (8 min read)
A forensic account of how United Wholesale Mortgage lost $600 million hedging a deal that had already fallen apart, and what earnings call disclosures to watch for after an unexplained loss. (11 min read)
Japan's three mass-market suit retailers, AOKi, Aoyama, and Konaka, are watching remote work hollow out their core business: this piece tracks each company's response and which strategies are gaining ground. (5 min read)
Acid Investments on Global Tax Free, Korea's leading VAT refund operator at tourist shopping districts: 41.5% return on invested capital last year, and a Japan joint venture launching in November. (10 min read)
The Oak Bloke's Harbour Energy 1H26 result: the cash generation, the debt trajectory, and whether the production decline rate changes the investment thesis. (13 min read)
Rijnberk InvestInsights on Uber's valuation problem: an extraordinary business priced as if growth is infinite and competition imaginary. (26 min read)
The Finance Corner on Nike's self-inflicted wound: down 75% from peak, insiders buying, and whether the brand damage is structural or fixable. (7 min read)
Asianometry delivers a competitive history of high-bandwidth memory in this watch: why SK Hynix leads, how hard the lead is to close, and what the next chip architecture changes. (38 min watch)
Chris Whalen on the first signs of private credit stress and how a large unwind would move through banks, insurers, and pension funds. Julia La Roche Show, in this watch. (31 min watch)
CFR's Spillover podcast on how China won Iran's energy crisis: the discount crude, the tanker fleet, and why the energy relationship is built to last, in this listen. (44 min listen)
Ben Thompson and Patrick O'Shaughnessy map who captures value as the AI boom moves from growth story to profit test. Winners & Losers of the AI Era covers big tech positioning, where China fits, and the framework for what happens when the spending cycle meets its first revenue headwind. A companion listen to the Nvidia piece above. (1 hr 27 min listen)
🤝 From Our Partners
A tool for writers and newsletter editors that reads every source you follow and scores each article against your standards, so you review the best out of what matters. Masthead.
Currently, I feature free content and collaborate with select paid newsletters. If you're a writer producing insightful investment analysis, whether free or behind a paywall, I'd love to discuss potential features. What would you like to see in future editions? Let me know in the comments.
Disclaimer: This newsletter is for informational purposes only and not investment advice. The intro reflects my views, while investment summaries are my interpretations of original authors' analyses. Information may not be fully verified and is subject to correction. Original authors' complete views may differ. Always do your own research before making investing decisions.
No posts

Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.