Specific, dated claims from across our letters and white papers, with what would actually prove each one right or wrong stated up front — reviewed on schedule, not just when it's convenient. This page is new, and most calls here are too recent to have resolved. That's the point: we're logging them now, while the outcome is still unknown, so nobody can accuse us later of only remembering the ones we got right.
Why every status below says "Pending" or "Watching." This publication is weeks old. A track record built on calls that haven't had time to resolve would be dishonest if it claimed anything more than that. We are not backfilling wins. Each entry below states exactly what would need to happen, and by when, for us to mark it correct or wrong — and we will update this page when that evidence exists, not before.
"The path to the trillion-dollar club runs through Veldhoven — and early 2027 is the NGE target." Market cap at time of writing: $733 billion, against a €38.8 billion order backlog.
Resolves when: ASML's market capitalisation crosses $1 trillion, or does not, by Q1 2027. Review date: March 2027.
"Germany is the cheapest G7 market for a reason, and this reason is not permanent." Called at a 55-60% Buffett Indicator and a 14x earnings multiple, the cheapest in the G7 at time of writing.
Resolves when: the DAX's valuation gap to G7 peers narrows meaningfully (relative re-rating), or the discount persists unchanged, over a 24-month window. Review date: June 2028.
"The valuation spread is the widest in a decade. The long-horizon investor ignoring 14 other markets because one dominates the news is leaving compounding on the table." US at 25-30x against Hong Kong at 10x and Germany at 14x.
Resolves when: undervalued markets (Hong Kong, Germany, UK) deliver superior total returns to the S&P 500 over a 3-5 year horizon from July 2026, or fail to. Review date: July 2029 (interim check July 2027).
Cited and adopted as house view: Epoch AI's estimate of "80% certainty" that high-quality public text-token data for AI training exhausts sometime in the 2026-2028 window.
Resolves when: major AI labs publicly confirm (or credibly avoid) a text-data scarcity constraint on frontier model training within that window. Review date: December 2028.
"The Airtel ending is the single most important unresolved question in the letter — genuine product appeal proven, durable monetized demand NOT proven." Flagged at the moment Airtel declined to renew its free-Perplexity-Pro partnership in favour of Adobe.
Resolves when: Perplexity secures a comparable-scale distribution partnership (telecom, OEM, or platform) within 12 months of the Airtel deal's end, demonstrating durable monetized demand independent of one partner — or fails to. Review date: January 2027.
244,000 US manufacturing jobs reshored in 2024 alone, against a persistent, named bottleneck of 2.1 million unfilled manufacturing jobs projected by 2030.
Resolves when: 2030 US Bureau of Labor Statistics data confirms or refutes the 2.1 million figure. Review date: Early 2031 (interim checks each year data is released).
15% effective annual global growth recommended as the actual planning base case (Section 2.4), against a 12% conservative floor and a 17-20% optimistic-to-stretch range — with fusion commercialisation timing named as the single largest source of uncertainty in every later-year figure.
Resolves when: White Paper 5, Part B — already published as a scenario exercise — is re-run with real 2028 data instead of simulated trends. Review date: mid-2028, per the commitment made in Letter 130 and White Paper 5, Part B itself.
Kevin Warsh's confirmation as Fed Chair framed as a hawkish turn — "forward guidance ended" — with implications named for the dollar, India's rupee, and the oil import bill.
Resolves when: the Fed's actual rate path through end-2026 confirms or contradicts a sustained hawkish stance relative to what markets priced in at the May 2026 confirmation. Review date: January 2027.
The claim that the next platform war is fought over which company AI purchasing agents trust, not over human attention — naming Perplexity's Comet, Stripe's agentic commerce protocol work, and OpenAI's Operator as early, unresolved competitors for that position.
Resolves when: a clear leader in agent-trust infrastructure emerges, or the category fails to materialise as described. This is a multi-year structural call with no near-term resolution point — marked "Too Early" rather than "Watching" for that reason. Review date: 2029, minimum.