Research records

Findings, then position.

Every record separates what the desk found from what one person concludes from it. Records in progress show their status openly rather than being hidden until publication.

Record 001 · 28 July 2026 Published

Subject — The desk itself

Why this desk exists, and what it will refuse to do

Coverage of Asian onchain capital markets is dominated by two sources: project communications, and Western analysts working from English-language summaries. Both under-read the primary record. The gap isn't insight — it's attention paid to filings, disclosure documents, and local-language releases at the moment they publish rather than three weeks later.

Position

The edge here is not being smarter. It is reading the source documents on the day they appear, in the language they appear in, and being willing to say when the announced thing and the built thing are different.

I would rather publish a narrow claim I can defend than a broad one I can't.

— Pratyush, EarlyLayers · 28 Jul 2026

Record 002 · In progress Drafting

Subject — Strium / Startale Group

A landscape primer: what Strium is actually building, from the primary record

A ground-up read of Strium's public materials, technical documentation, and the Startale Group relationship — separating architecture that ships from architecture that is described. Intended as the reference document a fund analyst can hand to an IC without further translation.

Position — pending

Held until the findings are complete. The desk does not publish a view ahead of the work that justifies it.

Record 003 · Queued In research

Subject — Tokenized securities stack

Comparing the tokenized securities stack across Asian venues

A side-by-side of issuance, custody, transfer, and settlement across the venues claiming tokenized securities capability in Japan, Korea, and Singapore — including which regulatory permissions each actually holds today.

Position — pending

Held until the findings are complete. The desk does not publish a view ahead of the work that justifies it.

Record 004 · 29 July 2026 · 2 primary sources Published

Subject — SBI Holdings / Startale Group

The distribution thesis is becoming a capital allocation thesis

SBI Group's approximately US$50 million investment in Startale Group changes the nature of the relationship. What was a distribution partnership around tokenized securities infrastructure now spans issuance, settlement, and distribution — with SBI holding direct economic exposure to the infrastructure it intends to distribute.

Position

The SBI investment makes the distribution thesis materially more credible.

Distribution partnerships are common in digital assets. Equity ownership is not. A commercial agreement can be replaced with relatively little cost; an equity stake creates incentives that are harder to unwind. It does not guarantee adoption, but it changes how I interpret SBI's commitment.

I no longer see Strium as a network merely seeking distribution through SBI. I see it as infrastructure SBI has begun integrating into a broader strategy spanning issuance, settlement, and distribution.

That changes the question. The issue is not whether SBI can introduce Strium to the market. The issue is whether combining infrastructure, regulated settlement assets such as JPYSC, and SBI's financial businesses creates a distribution advantage that competing tokenization platforms cannot easily reproduce.

The evidence that would change my mind is straightforward. By 31 December 2028, I would expect to see regulated issuance, institutional customers, or measurable product adoption that demonstrates the investment has translated into commercial execution. If those signals fail to emerge — or if competing platforms achieve comparable distribution without similar ownership structures — then the case for vertical integration weakens significantly.

For now, the public record supports a stronger conclusion than it did before the investment. SBI is no longer simply partnering with the ecosystem; it has become economically exposed to its success. That raises the evidentiary bar for anyone arguing the relationship is merely promotional.

— Pratyush, EarlyLayers · 29 Jul 2026

Record 005 · 4 August 2026 · 9 primary sources Published

Subject — SBI Holdings — the integration thesis, revised

Announced isn't built

Record 004 described SBI's tokenization activity as a vertically integrated approach. The documentary record as of July 2026 supported less: a small operating security-token market, a proof-of-concept L1, a live yen stablecoin whose settlement role had been tested only with a stand-in token on testnet, and a partnership announced in aspirational language. This record revises the characterization and documents how the desk came to publish it — the phrase was Startale's own, adopted without testing against the record. The revision is about our method as much as SBI's strategy.

Position

1. What stands. The Record 004 position and its 31 December 2028 criteria — regulated issuance, institutional customers, measurable adoption — are unchanged. Nothing in this record tested them. The revision is to the description, not the position.

2. What doesn't. The architectural description. As of July 2026 the documentary record supported components at different maturities — one operating market, a licensed second venue of unconfirmed status, a proof-of-concept L1, a stablecoin whose settlement role had been demonstrated only with a stand-in token on testnet, and an announced partnership — not an integrated stack. "Vertical integration" was the subject's own phrase for its intent. We published it as our finding.

3. The two claims, separated.

The methods claim: Record 004 promoted an announced strategy into an operating architecture. This is settled by the chronology as it stood when Record 004 was written. No future outcome bears on it — convergence in 2028 would vindicate SBI's strategy, not Record 004's description of it. No date.

The forward claim: Through 31 December 2028, I expect these components to keep developing as separate systems connected commercially — common ownership, partnership announcements, settlement pilots — rather than consolidating into a single production architecture. Falsifier: verifiable evidence by 31 December 2028 that ODX/START, Strium, and JPYSC settlement operate in production on a shared issuance or settlement layer, with the Ondo channel distributing assets issued on it.

— Pratyush, EarlyLayers · 4 Aug 2026