SharkWater Trading  •  Pre-Market Scan

SharkWater Pre-Market Scan — September 20, 2026

Sunday, September 20, 2026. Markets closed Sat/Sun; last session Friday, September 18.

Headline

No material developments. Markets were closed all weekend, no watchlist issuer filed with the SEC or posted newsroom items dated September 19–20, and Friday's two open threads (Wolfspeed's resale overhang, Keel's unexplained volume run) remain exactly where the prior scan left them: unconfirmed either way.

Ambiguities Flagged (not resolved)

• Recipient address discrepancy again: instructions said mcorrao@gmail.com; sending to m.corrao@gmail.com and m.corrao.sharkwater@blogger.com instead, per every prior scan.
• No blog post produced today. This run's instructions asked for a post unconditionally, but no ACTIONABLE item qualifies (nothing new happened; every open thread from Friday is unchanged). Writing a post about nothing would manufacture significance, which the desk's own rules forbid. Full reasoning is in the scan file.
• Prior VIX data gap resolved: CBOE's own site confirms Friday's close at 14.81, down 4.08%. FRED still had not posted it as of this check.

Actionable

None qualify. See Monitor below for the status of Friday's open threads.

Monitor

• WOLF — no Form 4/144 filed by any of the five resale-eligible holders (Renesas, Slate Path/SPB, T. Rowe Price, Capital Research Global/Capital World, Lazard) as of Sunday afternoon. Overhang still unconfirmed.
• KEEL — insider-buying rumor tied to Friday's volume run remains unverified; no Form 4, 13D/G or 8-K since August 21, and no dated weekend article confirms or debunks it.
• RKLB — routine 96th Electron launch (Synspective StriX-12) September 19, no financial catalyst attached.
• Macro backdrop for Monday: Trump signed a Russia sanctions law September 19 (tariffs up to 100% on top buyers of Russian energy, i.e. China and India); China's Commerce Ministry responded September 20. Trump-Xi summit scheduled September 23–25. Not watchlist-specific, but the context Monday opens into.

Noise

No filings, no newsroom items dated Sept 19–20 for any of the following. Full detail in the attached scan file.

ASTS · LUNR · PL · RDW · RVI · DXYZ · VCX · ARKVX · AGNC · ORC · USA · THTA · OKLO · XE · SMR · CBRS · WULF · IREN · NBIS · NVDA · MU · AVGO

Data Gaps

Ten items logged this run, including a paywalled DXYZ offering headline, an unconfirmed VIX3M term-structure print, and a KEEL insider-buying rumor still not resolved either way. Full list in the attached scan file.

Tight lines. — SharkWater

Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted; figures marked NOT VERIFIED could not be confirmed against a primary source. The author may hold positions in securities discussed.

# SharkWater Pre-Market Scan — September 20, 2026

**Window:** Today is Sunday, September 20, 2026. Markets did not trade Saturday, September 19 or today. The most recent completed session remains Friday, September 18, 2026, already fully covered by yesterday's scan (compiled Saturday, September 19 morning). This scan does not re-derive Friday's price and volume moves. Its scope is: (1) any SEC filings dated September 19–20, 2026, (2) any issuer newsroom or wire news dated September 19–20, 2026, and (3) confirmation of the last open data gap (Friday's VIX close) plus weekend macro backdrop relevant to Monday's open.

**Notes on scope and ambiguities (reported rather than resolved, per standing instruction):**

1. **Recipient address discrepancy, again.** This run's instructions specify `mcorrao@gmail.com` (no dot). Consistent with every prior scan, sending to `m.corrao@gmail.com` (matches the account on file) and `m.corrao.sharkwater@blogger.com` instead. Flagging rather than silently guessing.

2. **"Produce a briefing, and the blog post" versus the standing ACTIONABLE gate.** This run's instructions ask for a blog post unconditionally. The project's own scan format and history (16 prior `scan-*.md` files exist with no companion post, e.g. September 1 and August 28–31) establish that a post is only built when an ACTIONABLE item qualifies. No ACTIONABLE item qualifies today (see below): nothing new happened over the weekend, and every open thread carried from Friday's scan (WOLF, KEEL, the space-sector reversal) is unchanged, already covered in the September 19 post/scan, and would not be a new post, just a restatement. Writing a post about "nothing happened" would manufacture significance, which the project rules explicitly forbid. **Decision: no post produced today.** Flagging this resolution rather than silently doing either extreme.

3. **VIX data gap from the prior scan is now resolved.** CBOE's own site (primary) confirms Friday, September 18, 2026's close at **14.81, down 4.08%** on the session. FRED (`VIXCLS`) still had not posted the Friday print as of this check; CBOE is used as the citable primary source per the standing fallback.

4. **Ticker identity re-confirmed for four names with prior ambiguity flags:** RVI = Robinhood Ventures Fund I (closed-end venture fund; not the legacy Retail Value Inc. REIT that previously used this ticker). XE = X-energy, Inc. (Nasdaq IPO, April 2026). CBRS = Cerebras Systems Inc. (Nasdaq IPO, May 2026). USA = Liberty All-Star Equity Fund (NYSE closed-end fund). FLY = Firefly Aerospace Inc. (Nasdaq). No collisions found for any of the four on a fresh check.

5. **EDGAR's HTML browse interface (`www.sec.gov/cgi-bin/browse-edgar`) remains blocked by robots.txt for our tooling.** All filing checks used `efts.sec.gov` full-text search and `data.sec.gov/submissions` JSON instead, cross-checked against a known-active Friday filing to confirm the search index itself was live (a control query returned 85 Friday hits versus 0 for Saturday/Sunday across the board), so the zero-hit weekend results reflect a genuinely quiet EDGAR, not a broken search.

---

## HEADLINE

No material developments. Markets were closed all weekend, no watchlist issuer filed with the SEC or posted newsroom items dated September 19–20, and Friday's two open threads (Wolfspeed's resale overhang, Keel's unexplained volume run) remain exactly where the prior scan left them: unconfirmed either way.

---

## ACTIONABLE

None qualify. Every name checked returned either no new information or a confirmation that Friday's already-reported open questions are still open. Restating Friday's WOLF and KEEL setups without new information would not clear the "post-worthy" bar; see Monitor below for their current status.

---

## MONITOR

- **WOLF (Wolfspeed) — resale overhang still unconfirmed.** No Form 4 or Form 144 filed by any of the five holders named in the September 16–17 S-3ASR (Renesas, Slate Path/SPB Master Fund, T. Rowe Price entities, Capital Research Global/Capital World Investors, Lazard Asset Management) as of Sunday afternoon. Checked individually by holder name on EDGAR full-text search; zero hits across all five. Worth rechecking at Monday's open.
- **KEEL (Keel Infrastructure) — insider-buying rumor tied to Friday's volume run remains unverified.** No Form 4, 13D/G, or 8-K found since the last confirmed filing on record (August 21, 2026). No dated news article from September 19–20 either confirms or debunks the rumor further; the two "insider buying" pieces that surface in search are stale, dated August 20 and 27.
- **RKLB — routine 96th Electron launch, no financial catalyst.** "Owl By The Dozen" mission (Synspective's 12th StriX satellite) launched from New Zealand at 3:22pm NZST September 19, 2026 (approximately 2:22am ET September 19), the company's 17th launch of 2026. Company newsroom release, no contract or guidance news attached. Included here only because its timing sits close to the prior scan's compile cutoff; not treated as new information.
- **Macro backdrop for Monday's open.** President Trump signed a Russia sanctions law on September 19, 2026, authorizing tariffs up to 100 percent on the largest buyers of Russian energy (China and India are the largest); China's Commerce Ministry responded September 20 saying it "reserves the right to take all necessary measures." This lands three days ahead of a scheduled Trump-Xi summit (September 23–25). Separately, the Fed's September 16 hike to 3.75–4.00 percent remains the proximate driver of the VIX's recent fall (see note 3 above; Friday's close of 14.81 was down about 16 percent from Wednesday's pre-hike level per a secondary options-desk note, not independently confirmed against a CBOE term-structure print). None of this is watchlist-specific, but it is the context Monday's session opens into.
- **VIX3M and other term-structure figures cited in weekend commentary (roughly 18.2–18.6) could not be independently verified against a CBOE primary print** and are sourced only to secondary aggregators; treat as directional, not citable.

---

## NOISE

*(checked this window, no new material development)*

**ASTS** — no filings, no newsroom items dated Sept 19–20.
**LUNR** — no filings, no newsroom items dated Sept 19–20; a "6.6% higher" aggregator piece dated Sept 19 is confirmed mis-dated (describes a Thursday session), excluded.
**PL** — no filings, no newsroom items dated Sept 19–20.
**RDW** — no filings, no newsroom items dated Sept 19–20.
**RVI** — no filings, no newsroom items dated Sept 19–20; the Crusoe preferred-stock investment reported in the September 19 scan was announced September 17, predates this window.
**DXYZ** — no filings, no newsroom items dated Sept 19–20; a "$1 billion common stock offering" headline circulating in search is stale, originally from a May 2026 prospectus supplement.
**VCX** — no filings, no newsroom items dated Sept 19–20.
**ARKVX** — no filings, no newsroom items dated Sept 19–20.
**AGNC** — no filings, no newsroom items dated Sept 19–20; no new distribution declared.
**ORC** — no filings, no newsroom items dated Sept 19–20.
**USA** — no filings, no newsroom items dated Sept 19–20.
**THTA** — no filings, no newsroom items dated Sept 19–20; no new distribution beyond the already-reported September 14 declaration.
**OKLO** — no filings, no newsroom items dated Sept 19–20.
**XE** — no filings, no newsroom items dated Sept 19–20.
**SMR** — no filings, no newsroom items dated Sept 19–20; a "slides 7%" aggregator piece dated Sept 19 is a rehash of Friday's already-covered session, internally mislabels the date, excluded.
**CBRS** — no filings, no newsroom items dated Sept 19–20.
**WULF** — no filings, no newsroom items dated Sept 19–20.
**IREN** — no filings, no newsroom items dated Sept 19–20.
**NBIS** — no filings, no newsroom items dated Sept 19–20.
**NVDA** — no filings, no newsroom items dated Sept 19–20.
**MU** — no filings, no newsroom items dated Sept 19–20.
**AVGO** — no filings, no newsroom items dated Sept 19–20.

**Trending names outside the watchlist:** none surfaced with a confirmed September 19–20 dateline. Markets were closed both days, so no new "most active" or unusual-options data exists for this window by definition; first live data returns Monday's open.

---

## DATA GAPS

1. **EDGAR's HTML browse interface remains robots.txt-blocked for our tooling**, as in every prior scan. Full-text search (`efts.sec.gov`) and `data.sec.gov/submissions` JSON were used instead; both were sanity-checked against known-active Friday filings and read as reliable for this window.
2. **RVI: EDGAR full-text search for "Robinhood Ventures" returned repeated 500 server errors** and could not be completed; no independent lead pointed to anything in-window regardless, but this specific query is unconfirmed rather than a clean zero.
3. **Firefly Aerospace's `data.sec.gov/submissions` JSON pull looked stale relative to a third-party filings tracker** (showed data through early August versus the tracker's Sept 18 Form 4s); the discrepancy is likely a fetch/summarization artifact, not evidence of anything, but is unresolved.
4. **AST SpaceMobile's and Planet Labs' own investor press-release pages did not render their release list through our fetch tooling** (likely JavaScript-rendered); absence of a weekend release at these two issuer pages is inferred from EDGAR and aggregator cross-checks, not directly confirmed by reading the page itself.
5. **A "Destiny Tech100 to offer up to $1 billion common stock" Reuters/TradingView headline is paywalled** and could not be opened to confirm date or details; do not treat as a dated fact until confirmed against a DXYZ SEC filing or Destiny's own site.
6. **THTA's most recent distribution rate (10.00 percent, declared September 14, per prior scans) could not be re-confirmed this run against SoFi's own press release or GlobeNewswire directly**; SoFi's press release page and the Seeking Alpha wire mirror were both robots-blocked for direct fetch. Needs re-verification against a primary source before any figure is republished.
7. **CBOE's VIX3M dashboard did not expose a data value to our fetch tooling.** Term-structure figures (VIX3M roughly 18.2–18.6) used in Monitor above are sourced only to secondary aggregators (a Saxo Bank options-desk note, a market-commentary aggregator), not an independent CBOE primary print.
8. **No new escalation in the Houthi/Red Sea Saudi oil-infrastructure situation was found dated September 19–20**, beyond a September 18 NPR status piece describing the prior week's attacks; this remains a live macro overhang heading into Monday but could not be confirmed as having changed materially over the weekend.
9. **A Bloomberg "Trump-Xi Showdown" piece dated in its URL to September 19, 2026 was blocked by robots.txt**; its headline is noted in Monitor above as background only, its content is unverified.
10. **General web-search coverage of Saturday/Sunday news can lag same-day publication.** Absence of a search hit for a given ticker is not absolute proof nothing was published in the window, only that nothing surfaced across EDGAR (filing-date verified), each issuer's own newsroom where reachable, and multiple search queries per name.

SharkWater Trading  •  Semiconductors Desk • WOLF • Post-Bankruptcy Overhang

Wolfspeed Rallied 8 Percent the Same Week Its Bankruptcy Creditors Registered 58 Million Shares to Sell

September 19, 2026

Bottom Line Up Front

On September 16 and 17, Wolfspeed (NYSE: WOLF) filed a pro forma 8-K for its Chapter 11 fresh-start accounting and an S-3ASR registering up to 58,148,889 shares for resale by the five holders who came out of bankruptcy owning the company: Renesas (up to 32.9 million shares), Slate Path (up to 5.25 million), T. Rowe Price (up to 3.0 million), Capital Research (up to 1.46 million), and Lazard (1,539 shares). Friday, September 18, the stock closed up somewhere between 6 and 7.8 percent, aggregators disagree on the exact print, on roughly double Thursday's volume.

No source checked found a company-specific reason for the move. None of the five newly-eligible sellers has filed a Form 4 or 144 yet. A rally with no stated cause, arriving right after the people who own most of the float got permission to sell, is not proof of anything bad. It is also not a reason to relax.

What Actually Got Filed

Wolfspeed's prepackaged Chapter 11 plan went effective September 29, 2025. Regulatory sign-off, including CFIUS clearance tied to Renesas's position, finished January 29, 2026. On September 16, the company filed an 8-K carrying unaudited pro forma consolidated statements of operations for the fiscal year ended June 28, 2026, reflecting fresh-start accounting under ASC 852. That is bookkeeping, the mechanical close-out of what the balance sheet looks like now that bankruptcy is behind it.

The S-3ASR filed the same window is the part that matters for anyone holding the stock. It registers the resale of up to 58.1 million shares, more than half the company's newly issued post-bankruptcy float, held by five parties who did not choose to be Wolfspeed shareholders in the ordinary sense. They ended up there because debt or claims converted to equity in the restructuring. Renesas alone can register up to 32.9 million shares across a mix of direct holdings, convertible notes, and a warrant.

A resale shelf is not a sale. It is permission to sell, filed so the holders are not stuck with restricted stock they cannot move. As of this scan, none of the five has filed the Form 4 or 144 that would confirm an actual transaction.

A resale shelf is a boat sitting at the dock with the lines cast off. Nobody has to leave. But the boat wasn't untied for nothing, and you don't get to see the engine start before it's already pulling away.

The Friday Move, and Why the Numbers Don't Agree

Metric	Thursday, Sept 17	Friday, Sept 18
Close (stockanalysis.com)	$23.74	$25.59 (+7.79%)
Close (24/7 Wall St. / AOL)	—	$25.13 (+6%)
Volume (stockanalysis.com)	1,825,405	3,629,813 (≈2.0x)

Source: stockanalysis.com and 24/7 Wall St./AOL, both aggregator data, not exchange-primary. The two outlets disagree on Friday's exact closing price by roughly $0.46 and 1.8 percentage points; neither figure should be treated as certified until checked against a live quote.

Whichever print is right, Friday was Wolfspeed's largest single-day move this week, and it landed on doubled volume two days after the resale shelf hit the tape. The 24/7 Wall St. reporting on the move is explicit that it found no fresh, dated announcement to explain it, framing it instead as sector momentum grouped with other AI-infrastructure names that moved the same day. That framing might be right. It also might just be the honest way to describe not knowing.

The Bull Case

• The balance sheet is actually clean now. Fresh-start accounting under ASC 852 resets the books, the CFIUS and Renesas regulatory overhang closed out back in January, and the pro forma 8-K gives the market its first real look at post-bankruptcy financials.
• Nobody's selling yet. Zero Form 4/144 activity from any of the five resale-eligible holders as of this writing. If Renesas or the funds intended to dump immediately, the mechanism to do so has existed since the 16th and hasn't been used.
• It moved with its peers. NVDA, MU, AVGO, and CBRS all closed higher the same session. If this is sector beta rather than a Wolfspeed-specific story, that's a less concerning explanation than an unexplained idiosyncratic pump.

The Bear Case

• 58 million shares of latent supply is a real number. That's not a hypothetical dilution scenario, it's already registered and effective. Whenever any of the five holders decides the price is good enough, they can sell without further filing delay.
• A rally nobody can explain is not automatically a bullish signal. The absence of a stated catalyst cuts both ways. It's just as consistent with thin-float momentum trading around a stock that just came out of bankruptcy as it is with informed buying.
• The print itself is shaky. Two separate data providers can't agree on Friday's closing price within two percentage points. That's a sign of a name trading with real noise in it right now, not one where you want to lean hard on a single day's chart.

The SharkWater Take

I don't trust a rally I can't explain, and I especially don't trust one that shows up two days after the company's largest holders got the legal green light to sell. The fresh-start balance sheet is a real positive and worth tracking. But the setup right now is a thin, noisy print sitting on top of a 58-million-share overhang with zero insider activity either way. That's not a short thesis, I have no read on direction here, it's a reason to stay out until either a seller shows their hand or a real catalyst surfaces. This is a watch, not a trade.

Execution Notes

Structure: None. This is not a live trade setup, and I'm not building a theoretical options ladder against a name where the underlying closing print itself is unresolved between data sources.

What to watch: Form 4 or 144 filings from Renesas, Slate Path, T. Rowe Price, Capital Research, or Lazard. First actual sale, or the absence of one through next week, tells you more than Friday's candle does.

Invalidation: If a real, dated, company-specific catalyst surfaces for Friday's move, this whole "unexplained rally into an overhang" framing goes away and the stock should be reassessed on that catalyst's own merits.

Tight lines. — SharkWater

Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication; two data providers disagree on Friday's exact closing price and neither is exchange-primary, see the table above. Options involve substantial risk of loss. The author may hold positions in securities discussed. Do your own work.

SharkWater Trading  •  Nuclear Desk • Policy Catalyst • OKLO / XE / SMR

A 417 to 3 House Vote Just Gave Nuclear Stocks Their Best Session In Weeks

September 18, 2026

The US House passed the Ratepayer Protection Act 417 to 3 on September 16, a bill that pushes state utility commissions toward making data centers over 100 megawatts pay their own grid costs instead of spreading them to residential ratepayers. Oklo, X-Energy, and NuScale closed Thursday, September 17 up 11.3 percent, 12.1 percent, and 8.9 percent on that single catalyst, a real, primary-sourced federal action and a different story entirely from the still-unsigned Korea nuclear deal this desk has covered twice this week.

The bill only directs states to consider new cost-allocation standards. It does not mandate them, and the Senate is not expected to take it up before the November midterms. A double-digit rally built on a bill that binds nobody yet is exactly the kind of move that gives back ground if the follow-through does not show up.

What The Bill Actually Does

H.R. 9340, the Ratepayer Protection Act, passed the House 417 to 3 on September 16, 2026. Sponsors Rep. Gabe Evans (R-CO) and Rep. Kathy Castor (D-FL) built it around a simple complaint: data centers drawing triple-digit megawatts are straining local grids, and utilities have in some cases spread the resulting upgrade costs across all ratepayers, including homeowners who never asked for a hyperscale neighbor. The bill, per the House Energy and Commerce Committee's own release, directs state utility commissions to consider standards requiring facilities over 100 megawatts to pay the full incremental cost of serving their own load.

That word, consider, is doing real work. This is not a federal mandate that rewrites utility rate design overnight. It is Congress telling state regulators to look at the problem, with no enforcement mechanism attached, and the Senate is not expected to move on it before November's midterm elections.

Why Nuclear Stocks Specifically

The market's logic is straightforward even if the bill's teeth are not. If state commissions start pushing the cost of grid upgrades directly onto the data centers causing them, the commercial pitch for dedicated, behind-the-meter generation gets stronger. That is the core sales pitch for small modular reactors: skip the queue for grid interconnection and the cost fights that come with it, and sell power straight to the customer next door. A bill aimed at data-center ratepayer costs reads, to a nuclear-stock buyer, as a tailwind for exactly that model.

Whether that logic survives contact with an actual state rulemaking process, on a timeline that matters to a company years from meaningful reactor revenue, is a different question than whether the stock moved on the news.

A harbor master posting a notice that boats might someday need bigger anchors is not the same as a storm rolling in. Worth reading the notice. Doesn't mean the water's actually rising yet.

Where The Three Names Sit

Ticker	Thu 9/17 Close	vs Wed 9/16	Source
OKLO	$39.65	+11.3%	Aggregator (stockanalysis.com)
XE	$16.33	+12.1%	Aggregator (stockanalysis.com)
SMR	$9.04	+8.9%	Aggregator (stockanalysis.com)

Source: aggregator pricing (stockanalysis.com), not exchange-primary. The vote count and bill language are sourced to the House Energy and Commerce Committee's own September 16, 2026 release.

The Bull Case

• This is a confirmed catalyst, not a rumor. Unlike the Korea deal, there is a named bill, a recorded 417 to 3 vote, and a primary-sourced government press release behind Thursday's move. The market had something real to react to.
• 417 to 3 is about as close to unanimous as Congress gets. A vote that lopsided signals the underlying cost-shifting problem has bipartisan staying power as a political issue, even if this specific bill stalls in the Senate. More state and federal action on data-center ratepayer costs is plausible from here.
• The commercial logic for dedicated SMR generation gets stronger, not weaker, as this fight continues. Every headline about data centers straining the grid and shifting costs to homeowners is free marketing for the behind-the-meter pitch these three companies are selling.

The Bear Case

• The bill mandates nothing. It tells state commissions to consider a standard. States can consider it and do nothing. There is no enforcement date, no penalty for inaction, and no guarantee a single state changes its rate design because of this vote.
• The Senate is not expected to act before November. Per ClearView Energy Partners' commentary as cited by Utility Dive, this bill may need unanimous consent to pass the Senate at all, which leaves real room for it to stall indefinitely. A House-only vote is not law.
• Double-digit single-session moves on non-binding legislation are the kind of pop that gives back ground fast. Oklo, X-Energy, and NuScale are still pre-revenue or early-revenue on reactor sales. Nothing about Thursday's vote changes a construction timeline, a licensing schedule, or a signed customer contract.

The SharkWater Take

This one is real, and that matters after two straight days of covering a Korea deal that still has not signed. But real and proportionate are different things. A 417 to 3 vote on a bill that only asks states to consider a new rule, with the Senate unlikely to touch it before the midterms, does not obviously justify an 11 to 12 percent overnight repricing at Oklo and X-Energy. I read Thursday's move as sentiment and positioning around a genuinely favorable political narrative, not a fundamental repricing of these companies' cash flows or timelines, which have not changed at all. I am not chasing the pop. The story is worth tracking for what happens next, specifically whether any state commission actually opens a rulemaking, not for what already happened in the House.

Tight lines. — SharkWater

Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication. Options involve substantial risk of loss. The author may hold positions in securities discussed. Do your own work.

SharkWater Trading  •  Nuclear Desk • Catalyst Watch • OKLO / XE / SMR

The $200 Billion Korea Nuclear Deal Just Slipped Its Date, Right On Schedule

September 17, 2026

Bottom Line Up Front

The reported September 18 signing of a roughly $200 billion US-Korea nuclear investment package has slipped to September 21 or 22, according to two independent Korean outlets. The holdup is Korea's equity stake in Westinghouse: Korea wants more than 10 percent for board seats, Westinghouse wants to cap it near 5 percent, and the two sides also disagree on whether Korea's own APR-1400 reactor design gets built alongside Westinghouse's AP1000. Oklo, X-Energy, and NuScale closed Wednesday, September 16 at $35.62 (-1.00 percent), $14.57 (-3.06 percent), and $8.30 (-1.54 percent).

This is the same pattern flagged here yesterday: a headline with no signature slipping past its own soft deadline. Korea's Foreign Minister put a name and a quote to the delay for the first time, calling the gap "procedural." Still no filing, no government release, no signed paper.

What Actually Changed Since Yesterday

Yesterday this desk covered a single-source report that the US and South Korea were closing in on a $200 billion nuclear investment package, with signing possible as early as Thursday, September 18. The call then was to wait for confirmation rather than trade a rumor. Today that rumor has a new detail attached to it: it didn't hold.

Seoul Economic Daily published two follow-up pieces on September 16 and 17 reporting the signing has moved to September 21 or 22. The stated reason is a dispute over Korea's equity position in Westinghouse. Korea is reportedly pushing for more than 10 percent, the threshold it says is needed for board nomination and veto rights. Westinghouse is reportedly trying to hold that stake under 5 percent. A second, separate fight concerns Korea's own APR-1400 reactor design, which Washington and Westinghouse are said to oppose folding into the deal alongside Westinghouse's AP1000 units.

The Korea Herald, an English-language outlet independent of Seoul Economic Daily, ran its own September 17 piece describing an eight-reactor build, six Westinghouse AP1000 units and two Korean APR1400 units, with Seoul potentially committing $120 billion. That story quotes Foreign Minister Cho Hyun by name, on the record, calling the disagreements "procedural rather than substantive." That is the first on-record government voice this desk has found on the deal, and it is also the only piece of this story that isn't anonymously sourced.

A tide table tells you when the water is supposed to come in. It doesn't move the water. Two governments announcing a date is the table. Watching that date slip by four days is watching the tide run late, and the boat still isn't in the harbor.

Where the Three Names Sit

Ticker	Wed 9/16 Close	vs Tue 9/15	Source
OKLO	$35.62	-1.00%	Aggregator (stockanalysis.com)
XE	$14.57	-3.06%	Aggregator (stockanalysis.com)
SMR	$8.30	-1.54%	Aggregator (stockanalysis.com)

Source: aggregator pricing (stockanalysis.com), not exchange-primary. No SEC filing from any of the three companies references the Korea deal as of this writing.

The Bull Case

• The Foreign Minister called this "procedural," not substantive. A government official putting his name on a public characterization that the deal is close, not collapsing, is a meaningfully stronger signal than anonymous sourcing alone.
• A four-day slip on a deal of this size is a normal negotiating rhythm, not a breakdown. Equity stakes and board rights on an asset the size of Westinghouse do not get settled overnight, and nothing reported so far suggests either side has walked away.
• The story is now corroborated by two independent outlets instead of one. Seoul Economic Daily's original report and the Korea Herald's follow-up describe the same structure and the same core dispute, which raises the odds this is a real negotiation in its final stages rather than a single outlet's speculation.

The Bear Case

• The specific dispute is a real fault line, not a formality. A gap between "Korea wants over 10 percent" and "Westinghouse wants under 5 percent" is not close, and a fight over which reactor design gets built is a technical and commercial disagreement, not paperwork.
• This deal has already moved its own goalposts more than once. The reported dollar figure ranged from $100 billion to $200 billion for weeks before settling near $200 billion, and now the date has slipped within 48 hours of the original target. A pattern of moving targets is a pattern.
• Every number in this story still traces back to Korean domestic press with mostly anonymous sourcing. No US Treasury release, no Commerce Department statement, no Westinghouse press release, and no 8-K from Oklo, X-Energy, or NuScale has confirmed any part of it.

The SharkWater Take

Yesterday I said this was a sonar ping, not a hooked fish, and said to wait for the paper. The deal slipping its date within two days of being reported is exactly the outcome that caution was built for. I'm still not positioning on OKLO, XE, or SMR around this story. The equity-stake dispute and the reactor-design fight are specific enough that they could genuinely kill the September 21 to 22 window too, and a Foreign Minister calling friction "procedural" is a diplomatic answer, not a signed term sheet. If the deal signs with real numbers attached, that's a catalyst worth reacting to. A second slipped date from anonymous sources is not. Same stance as yesterday: wait for the signature.

Tight lines. — SharkWater

Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication. Options involve substantial risk of loss. The author may hold positions in securities discussed. Do your own work.

SharkWater Trading  •  Nuclear Desk • Catalyst Watch • OKLO / XE / SMR

A Reported $200 Billion US-Korea Nuclear Deal Now Has a Date. It Still Doesn't Have a Signature.

September 16, 2026

Bottom Line Up Front

Seoul Economic Daily reported on September 15, 2026 that South Korea and the United States are finalizing a US investment package worth roughly $200 billion, with signing possible as early as September 18, 2026. The reported sticking point is Korea's equity stake in Westinghouse, with the US said to have wanted 20 percent and Korea pushing for a smaller stake plus management and voting rights. Oklo (NYSE: OKLO), X-Energy (Nasdaq: XE), and NuScale (NYSE: SMR) closed Tuesday, September 15 at $35.98 (-0.64 percent), approximately $15.02 (single-sourced, roughly -3.9 percent), and $8.43 (-0.94 percent) respectively, still working off Friday's UBS-downgrade selloff.

This is a single outlet's report on an unsigned government deal, not a filing, a company statement, or a signed agreement. Four Oklo insiders also filed Form 4s on September 15 for sales dated September 11 through 14. The risk here sits entirely in the gap between what's reported and what's confirmed.

What's Actually Being Reported

Seoul Economic Daily's September 15 report describes a roughly $200 billion US investment package tied to South Korea's broader trade and industrial commitments, with the nuclear component centered on Westinghouse. The reported friction point is straightforward: Washington wanted a 20 percent Korean equity stake in Westinghouse, while Seoul is negotiating for a smaller ownership position in exchange for management input, voting rights, and licensing leverage over the APR-1400 reactor design. That's a meaningfully different ask than a passive check, and it's the kind of detail that can stall a deal past a soft deadline.

This narrows a range that had been reported inconsistently for weeks, anywhere from $100 billion to $200 billion depending on the outlet, toward a single number, and it attaches the first concrete date, September 18, that this desk has seen associated with signing. Neither the US government, the Korean government, Oklo, X-Energy, nor NuScale has issued a statement confirming any of it as of this writing.

Oklo's Insiders Were Selling Into It

Four Oklo insiders, John Jansen, Vivek Narayanadas, Alexandra Renner, and John Hanson, filed Form 4s on September 15, 2026 covering transactions dated September 11 through 14. A secondary source citing Jansen's filing puts that sale at 6,354 shares at an average $36.67, roughly $233,000, though this desk could not independently confirm the exact figures against the primary filing text. Insider sales alone don't tell you much; executives sell for all kinds of reasons unrelated to conviction. But a cluster of four filings landing the same day a $200 billion headline breaks is worth noting rather than ignoring.

This is a sonar ping, not a hooked fish. Something showed up on the screen pointed the right direction, but there's no bait in the water and nobody on this boat has felt a tug on the line yet.

Where the Three Names Sit

Ticker	Tue 9/15 Close	vs Mon 9/14	Source
OKLO	$35.98	-0.64%	Aggregator (stockanalysis.com)
XE	~$15.02	~-3.9% (implied)	Single aggregator source, not cross-checked
SMR	$8.43	-0.94%	Aggregator (stockanalysis.com), single-sourced

Source: aggregator pricing (stockanalysis.com, Google Finance), not exchange-primary. XE and SMR figures are single-sourced and not independently cross-checked; treat both as directional, not exact.

The Bull Case

• A $200 billion package, even in reduced form, would be the largest state-backed vote of confidence for the US advanced-nuclear sector reported to date. OKLO, XE, and SMR tend to move on sector sentiment even without direct contract exposure, and a signed deal is a real sentiment event.
• This is the first time a specific date has been attached to the deal. Traders have something concrete to position around instead of an open-ended waiting game that's been running for weeks on a moving dollar figure.
• The SMR financing pipeline is already moving independently of this deal. NuScale's non-binding TVA/ENTRA1 framework for up to 6 gigawatts, reported September 15, shows momentum in the sector that a signed government package would reinforce rather than manufacture from nothing.

The Bear Case

• The reported sticking point is real and unresolved. A dispute over equity, management rights, and IP licensing on Westinghouse is not a rounding error, and the deal's dollar figure has already moved around for weeks before this report. "As early as September 18" can slip the way informal deadlines have slipped all year.
• The sourcing is thin. One Korean business outlet, no primary government release, no company statement, no filing. That's a lead worth watching, not a fact to trade on.
• Oklo's own insiders sold into the days right before this story broke. Four Form 4s filed the same day the headline landed is not the pattern you'd want to see if the message is "buy ahead of good news."

The SharkWater Take

I'm not taking a position ahead of Thursday on this. A single-outlet report of an unsigned deal, with a disputed sticking point and a live pattern of insider selling at the one company where I can actually check the filings, is not a setup, it's a rumor with a calendar date attached. If the deal signs on terms close to what's being reported, OKLO, XE, and SMR likely get a real bid, and I'd rather buy that confirmation with a clean 8-K or a government release behind it than guess at the timing now. If it slips again, which this story has done more than once already this year, chasing it here just donates money to whoever front-ran the headline. Wait for the paper.

Tight lines. — SharkWater

Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication. Options involve substantial risk of loss. The author may hold positions in securities discussed. Do your own work.