SharkWater Trading  •  Digital Infrastructure • Power • Speculative

KEEL: A Two Billion Dollar Bet That Somebody Signs A Lease

August 18, 2026

Bottom Line Up Front

Keel Infrastructure is the company formerly known as Bitfarms. It redomiciled to the US, rebranded on April 1, and is converting itself from a Bitcoin miner into a landlord for AI and high performance computing. Market cap is roughly $2.1 billion at about $3.54 a share.

Revenue from the HPC business is zero. The legacy mining revenue that remains carried a gross margin of negative 285 percent last quarter. What you are actually buying is 2.2 gigawatts of pipeline, three permitted sites, $819 million of liquidity, and the proposition that a creditworthy tenant signs a long term lease. Everything else in the financials is noise on the way to that one event.

What You Are Actually Buying

Strip away the transformation story and Keel is three things.

One: interconnection queue position. A 2.2 gigawatt development pipeline with established grid interconnections already in place, spanning PJM in Pennsylvania, Grant County PUD in Washington State, and Hydro-Québec territory in Canada. In a market where the binding constraint on AI buildout is power delivery rather than chips, an existing interconnect is the scarce asset. CEO Ben Gagnon put it plainly on the Q2 call: power is the constraint, and everything else is downstream of it.

Two: a cash pile. $819 million of liquidity as of August 7, made up of roughly $698 million in unrestricted cash and $121 million in unencumbered Bitcoin, funded in part by a $458 million convertible note offering during the quarter.

Three: a melting ice cube that used to be the business. All US Bitcoin mining has been decommissioned. The remaining legacy operation produced $30.4 million of revenue in Q2 against $117.2 million of cost of revenues.

The Sites

Site	Location	Status
Moses Lake	Washington	First Vertiv modules delivered. Likely first to commission.
Panther Creek	Pennsylvania (PJM)	Conditional land development approval secured.
Sharon	Pennsylvania (PJM)	Zoning secured, long-lead items arriving.
Scrubgrass	Pennsylvania	Potential gigawatt. 750 MW load study with FirstEnergy, visibility expected Q3 to Q4.
Sherbrooke	Québec	Agreement with Hydro-Sherbrooke for conditional transfer of 96 MW plus land purchase.

Management describes active negotiations with multiple prospective tenants at each of the three priority sites, targets lease execution in 2026, and commissioning in 2027. Environmental permitting is progressing across all three. Fiber contracts are in final execution.

Note what is absent from that list: a signed lease.

The Q2 Numbers, And The One Line Everybody Skipped

Q2 2026 (continuing operations)	Amount	vs Q2 2025
Revenue	$30.4M	-50%
Cost of revenues	$117.2M	+81%
Gross margin	-285%	from -6%
G&A	$31.3M	+62%
Operating loss	-$140.8M	from +$10.8M
Depreciation & amortization	$84.1M	+218%
Gain on derivatives	+$77.0M	from +$3.8M
Loss from continuing ops	-$64.0M, or -$0.11/sh	from +$13.2M
Adjusted EBITDA	-$23.7M	from +$6.6M

Look at the derivative line. Keel lost $140.8 million at the operating level. It reported a loss from continuing operations of $64.0 million. The bridge between those two numbers is largely a $77.0 million non-cash gain on derivative assets and liabilities, tied to the marks on its convertible notes and the associated capped call transactions.

That gain is an accounting revaluation, not cash, and it reverses when the marks move the other way. Anyone reading the headline loss per share of eleven cents and concluding the burn is modest has read the wrong line. The operating burn is roughly twice that, and the adjusted EBITDA figure of negative $23.7 million is the honest read on cash profitability.

This is the same species of thing as the return of capital footnote in an income ETF or the customer warrant amortization at a newly public chip company. The interesting number is never the one in the headline. It is the one three lines below it that explains why the headline looks the way it does.

The Bitcoin Stack Is Bridge Financing, And It Is Nearly Spent

Between April 1 and August 7, Keel sold 1,085 Bitcoin for $75 million in proceeds, an average of roughly $69,100 per coin. The remaining balance stands at 1,861 BTC, carried at about $121 million.

Do the arithmetic on what that means. The Bitcoin treasury is now about 15 percent of total liquidity. The company started this transition with just under 2,500 coins and has been selling them down deliberately to fund the pivot. That funding source has maybe one more meaningful draw left in it.

From here, the capital comes from the convertible notes already raised, from future equity or debt, or from project level construction financing. And project financing is precisely what requires a signed lease with a creditworthy tenant.

Here is the structural point that makes this company legible. Management has said it intends to finance construction through investment grade tenants and credit wrapped leases rather than by tapping capital markets. That means the lease is not merely the revenue event. The lease is the key that unlocks the construction financing. The $819 million on the balance sheet is designed to carry the company to signature, not to build the data centers. Understand that and the entire investment case collapses into a single question with a binary answer.

What The Market Is Paying Per Megawatt

Conventional multiples are meaningless here. Price to sales on a business being deliberately shut down tells you nothing. So price the asset instead.

Basis	Capacity	Market cap per MW
Full stated pipeline	2,200 MW	~$0.95M
Active plus secured only	~771 MW	~$2.7M

The gap between those two rows is the entire debate. The 2.2 gigawatt figure includes roughly 1.5 gigawatts described as expansion and evaluation, which is a different category of asset than energized capacity with a live interconnect. If the Scrubgrass 750 megawatt load study with FirstEnergy comes back favorably in the Q3 to Q4 window, a large chunk of that speculative bucket moves toward the secured column, and the per megawatt math changes materially.

Also remember that roughly $819 million of the $2.1 billion market cap is liquidity sitting on the balance sheet. Back that out and the market is assigning something on the order of $1.3 billion to the pipeline, the permits, and the team, against convertible obligations that will eventually either convert into shares or come due.

Bull Case

Power is genuinely the bottleneck in AI infrastructure right now, and Keel holds interconnection positions in PJM and the Pacific Northwest that would take years and considerable litigation to replicate from scratch. The balance sheet is the strongest in company history by management's own account, funded partly with 1.25 percent convertible paper, which is remarkably cheap capital for a business at this stage. Three sites are near full permitting with multiple prospective tenants negotiating for each. The company brought in a President specifically to run commercial and expansion activity, which is what you do when you expect to be signing contracts. The legacy mining drag is now largely decommissioned rather than lingering. Analyst consensus sits at Strong Buy with an average target near $6.67, roughly 88 percent above the current price. And a single anchor lease at any one of these sites would re-rate the equity in a way no amount of quarterly progress reporting can.

Bear Case

There is no lease. "Active negotiations" and "deepening commercial engagement" are the language of companies that have not yet closed. Hyperscalers and neoclouds have many suitors, and Keel is a first-time HPC developer competing against operators with track records.

The burn is real and the runway has a shape. Negative $23.7 million adjusted EBITDA per quarter is the baseline before development capital. Miss the 2026 lease target and 2027 commissioning slips, which pushes the financing question into a market that may be less friendly to speculative data center credit than it is today.

Dilution is structural, not hypothetical. The company's own risk factors name potential dilution from future stock issuances, conversion of the convertible notes, and exercise of options and warrants, plus counterparty risk on the capped call transactions.

Execution risk is stacked. Environmental permitting across three jurisdictions, supply chain and tariff exposure on long-lead equipment, community opposition to data centers, and regulated power rates in Québec, Pennsylvania, and Washington. Any one of those can add quarters.

And the residual Bitcoin exposure cuts both ways. $121 million of the liquidity is a volatile asset the company is trying to sell into strength while also depending on it.

A Word About The 2x ETF

There is a Defiance Daily Target 2X Long KEEL ETF trading under KEEX, with about $4.7 million in net assets and a 1.31 percent expense ratio. Its 52 week range runs from $15.35 to $181.67.

That range is not a typo and it is not an opportunity. It is what daily-reset leverage does to a volatile underlying over time. The fund resets its 200 percent exposure every day, so a stock that chops sideways with big swings grinds the leveraged product down regardless of where the underlying finishes. The prospectus language is explicit that investors could lose their entire principal in a single trading day.

Worth knowing for a second reason: flows into and out of leveraged products like this can create mechanical buying and selling pressure in a $2 billion stock. Some of the daily moves in KEEL are not information. They are rebalancing.

The SharkWater Take

I like this setup more than I expected to, and I would size it like the venture position it is.

The thesis is unusually clean. Most speculative names ask you to believe five things at once. Keel asks you to believe one: that a creditworthy tenant signs a long term lease at Moses Lake, Panther Creek, or Sharon, on terms that make the construction financing work. If that happens, the pipeline reprices from optionality to contracted cash flow and the current $2.7 million per secured megawatt looks cheap. If it does not happen in 2026, you are holding a cash-burning developer with convertible obligations and a shrinking Bitcoin stack, waiting on a 2027 that costs more to reach.

What I would actually watch, in order: the Scrubgrass load study result in the Q3 to Q4 window, because it either validates or deflates two thirds of the stated pipeline. Then any 8-K announcing a lease, because that is the entire thesis in a single filing. Then the cash balance each quarter against the $23.7 million adjusted EBITDA baseline plus development spend, because runway is what buys negotiating leverage, and management explicitly framed the $819 million as the source of that leverage.

One thing I would not do here is sell puts the way I would on an asset-backed closed end fund. When you sell a put on a fund holding Treasuries, assignment hands you T-bills at a discount. When you sell a put on Keel, assignment hands you a pre-revenue developer with negative gross margins and a binary catalyst. The premium may look attractive because the implied volatility is enormous, and that volatility is telling you something true. Respect it.

This is a position you size in basis points, enter with a thesis you wrote down, and exit on the catalyst rather than the narrative. It is not a core holding and the company's own filings say trading in its securities should be considered highly speculative. I take management at its word on that.

Fair winds and following seas.

---

Disclaimer: This post is for informational and educational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any security. Financial figures are drawn from Keel Infrastructure's Q2 2026 results release and related SEC filings as of August 2026 and are subject to revision. Market capitalization, per-megawatt calculations, and capacity splits are the author's estimates derived from company disclosures, not reported company metrics, and depend on assumptions about share count and pipeline classification. Forward-looking statements regarding lease execution, permitting, and commissioning timelines are management's and are not assured. The company is loss-making, has negative gross margins in its legacy segment, and its own filings state that trading in its securities should be considered highly speculative. Leveraged ETFs referenced are not recommendations and carry risk of total loss. Verify all figures against primary filings before acting. Past performance does not guarantee future results. Consult a licensed financial professional regarding your specific situation.

 SharkWater Trading  •  AI Infrastructure • Semis • Counterparty Risk

ORCL, MRVL, CBRS: Three Floors Of The Same Building, And Two Of Them Have The Same Landlord

August 17, 2026

Bottom Line Up Front

Oracle rents the building. Marvell sells the wiring. Cerebras builds a specialized engine that goes inside. All three are levered to the AI capacity buildout, but they are not three separate bets.

Roughly half of Oracle's $638 billion backlog traces to OpenAI. Roughly eighty percent of Cerebras' $25.4 billion backlog traces to OpenAI. Owning both is not diversification. It is the same counterparty bet placed twice, at two very different valuations. Marvell is the odd one out, and that is the most interesting thing about it.

The Stack At A Glance

	ORCL	MRVL	CBRS
Role	Capacity landlord	Interconnect and custom silicon	Wafer-scale inference
Price (Aug 17)	~$147.57	~$234.20	~$251.98
Market cap	~$420B	~$199B	~$60B
Revenue base	$67.4B FY26	~$10.8B run rate	$880 to $890M FY26 guide
Roughly, price to sales	~6x	~18x	~68x
Backlog (RPO)	$638B	Not reported this way	$25.4B
Backlog vs revenue	~9.5x	n/a	~29x
From 52-week high	Down ~57%	Down ~29% from $329	Down ~35% from $386
Next catalyst	FQ1 2027, ~Sept 8 to 14	FQ2 2027, Aug 27	SUPERNOVA event

Read the price to sales row from left to right and you are reading an escalator of expectations. Same theme, three wildly different prices for exposure to it.

Oracle: A Great Business Attached To A Terrifying Cash Flow Statement

Fiscal Q4 2026 was, on the surface, one of the better quarters any megacap has printed. Total revenue $19.2 billion, up 21 percent. Cloud revenue $9.9 billion, up 47 percent. Infrastructure alone $5.8 billion, up 93 percent. Full year revenue $67.4 billion, up 17.4 percent, with net income up 36.5 percent.

And RPO, the contracted backlog, hit $638 billion, up 363 percent year over year and up roughly $85 billion in that quarter alone.

The stock is down about 57 percent from its high.

Why The Market Stopped Caring About The Backlog

Because backlog is a promise, and promises have to be financed before they become cash.

Metric	FY2026
Capital expenditure	~$55.6B (above the $50B target)
Free cash flow	-$23.7B
FY2027 capex guidance	Up to ~$95B, roughly $70B net Oracle spend
Corporate debt	Above $110B

For context, Oracle averaged about $1.7 billion in annual capex between fiscal 2018 and 2021. It is now spending that much roughly every eight days.

The catalyst chain that broke the stock was specific. December's Q2 print disclosed $12 billion of capex in a single quarter and a $15 billion raise to the full year forecast, and shares fell 15 percent on the day. Blue Owl Capital then walked away from a $10 billion data center financing deal, reportedly over Oracle's spending pace and debt load. S&P downgraded. Securities class actions followed. And reports surfaced that OpenAI, the counterparty behind an estimated $300 billion of that backlog, had missed internal revenue targets.

Management has an answer, and it is a decent one: much of the recent RPO growth came from contracts where the customer either prepaid for the GPUs or supplied the GPUs directly, which reduces what Oracle itself has to finance. That is a real mitigant and it deserves weight.

The right analogy is a contractor who has just signed the largest order book in the history of the trade, from a client whose own financing is not fully secured, and who must now borrow heavily to buy the materials. The order book is real. The question is who runs out of money first.

Bull: 44 analysts carry an average target near $246, roughly 68 percent above the current price. The underlying database and applications business still throws off cash. If even half of that backlog converts on schedule, the revenue math is transformative, and the stock is priced at roughly six times sales for it.

Bear: Every dollar of that backlog assumes the customer can pay and Oracle can build. Gross margin has already compressed from 81 percent to around 69 percent as revenue mix shifts toward infrastructure. Negative free cash flow of $23.7 billion in a single fiscal year, funded by debt, in a business where the largest customer is a company that is itself not yet reliably profitable, is the definition of a leveraged bet on somebody else's business plan.

Marvell: The Only One Here Selling To Everybody

Marvell is up roughly 161 percent in 2026, joined the S&P 500 in June, took a $2 billion investment from NVIDIA, and got publicly called a future trillion dollar company by Jensen Huang. Q1 FY2027 revenue was $2.42 billion, up 27.6 percent, with Q2 guided to about $2.7 billion. Roughly 75 percent of revenue is tied to data center and AI demand.

It also trades around 81 times trailing earnings and about 18 times sales, and it peaked at $329 in June before falling 35 percent and clawing back to the mid $230s. This is not a cheap stock and it is not a calm one.

The Narrative Gap Worth Watching

Most of the retail conversation treats Marvell as a custom AI silicon story, a Broadcom competitor. On that axis the position is honestly mediocre. Marvell holds an estimated 20 to 25 percent design share against Broadcom's roughly 70 percent, and Broadcom serves three of the five major hyperscaler programs including Google's TPU, Meta's Iris, and OpenAI's Jalapeño. Marvell has Amazon Trainium and Microsoft Maia 200.

But management has guided optical interconnect revenue to grow above 70 percent in fiscal 2027, versus 20 percent plus for custom silicon. In optical DSPs, Marvell holds an estimated 60 to 65 percent share, built on a position that predates the AI boom entirely. The Celestial AI acquisition added plasmonics-based silicon photonics and XConn added PCIe and CXL switching.

The company's own guidance says the growth engine is the interconnect business, not the ASIC business the market argues about. If you are underwriting this name, that is where the work should go.

Bull: Every AI cluster needs optics regardless of whose accelerator wins. Revenue is recognized on shipment, not promised over eight years. Marvell has no OpenAI counterparty concentration, and its custom silicon customers are Amazon and Microsoft, two of the best-capitalized buyers on earth. Analyst consensus is Strong Buy with an average target near $257.

Bear: 81 times trailing earnings after a 161 percent run leaves no margin for a soft quarter, and one downgrade to Hold already flagged exactly that. Operating margin sat at 14 percent in Q1, in line with a year prior, meaning the revenue growth has not yet produced the operating leverage the story promises. Marvell is materially smaller than Broadcom and NVIDIA and has to outspend its weight class to stay relevant. Earnings land August 27, one day after NVIDIA, in the middle of the most concentrated AI sentiment window of the year.

Cerebras: Read The Warrant Footnote

Cerebras went public in May 2026, raising $6.4 billion in the largest semiconductor IPO on record, oversubscribed roughly twenty times. It ran to $386.34 intraday, collapsed to $160.81 by late June, and now sits around $252. Reported beta is above 5. Treat every number in this section as attached to a stock that moves 15 percent on a Monday.

The Q2 print on August 12 was, operationally, excellent. Core revenue $210 million, up 103 percent, against consensus of $193.6 million. Cloud revenue nearly quadrupled. Full year core revenue guidance was raised to $880 to $890 million from $855 to $865 million, and full year gross margin guidance was raised to 41 to 43 percent from 38 to 41 percent.

The stock fell 13 to 20 percent.

Here Is Why, And Almost Nobody Explained It Properly

In December 2025 Cerebras signed a master relationship agreement with OpenAI covering 750 megawatts of inference capacity through 2028, valued above $20 billion, with an option for an additional 1.25 gigawatts by 2030. OpenAI also provided a roughly $1 billion working capital loan in January 2026 and received warrants for approximately 10 percent of the company.

Under ASC 606, warrants issued to a customer as part of a commercial arrangement are treated as consideration paid to that customer. The fair value sits on the balance sheet as a customer warrant asset and gets amortized against revenue as the customer buys compute.

As of June 30, Cerebras carried roughly $1.128 billion of customer warrant assets, split $167.8 million current and $960.6 million non-current, and the 10-Q states this amortizes through October 2031.

Translation: there is a permanent, structural wedge between the "core revenue" management guides to and the GAAP revenue that hits the income statement, and it runs for another five years. If your data feed compares management's core guidance to a GAAP consensus, you will get a phantom miss every single quarter. That is roughly what happened on August 12, and it produced a $13 million GAAP shortfall that knocked a fifth off the market cap on a quarter where the business beat and raised.

Then on August 17 the stock ran 15 percent on renewed attention to the OpenAI relationship. Same company, same footnote, opposite direction, five days apart.

Bull: Independently benchmarked as the fastest inference hardware available, with an anchor contract above $20 billion, AWS distribution through Bedrock, and over $10 billion of capital to deploy. RPO of $25.4 billion against an $885 million revenue year, with roughly 22 percent, about $5.59 billion, expected to convert within 24 months. Margin guidance was raised, not cut.

Bear: Roughly eighty percent of that backlog is one customer, and the rest is concentrated in AWS and G42. The company is not profitable. It trades near 68 times current year revenue. Any slippage in OpenAI's deployment schedule does not dent the story, it removes it. And OpenAI holds warrants on ten percent of the company, which is an alignment mechanism and a dilution overhang in the same instrument.

The SharkWater Take

The concentration point is the one I would tape to the monitor. If OpenAI is more than half of Oracle's backlog and roughly four fifths of Cerebras' backlog, then a position in both is one position in OpenAI's ability to fund what it has signed, expressed through two balance sheets you do not control. Anyone holding both and feeling diversified should size accordingly.

On the individual names:

Oracle is the only one of the three where the pessimism is already in the price. Down 57 percent, roughly six times sales, with a real underlying enterprise business underneath the AI wager. It is also the one with $110 billion of debt and negative $23.7 billion of free cash flow, so "cheap" is doing a lot of work. The September print is the tell, and the numbers that matter are cloud growth against the guided 57 to 63 percent range, capacity actually delivered, and any commentary on the credit outlook. Not the RPO headline.

Marvell is the cleanest business model of the three and the one I understand best. It ships product, books revenue, and gets paid, with no eight-year backlog requiring an act of faith. It is also priced for perfection at 81 times earnings after a 161 percent year, reporting the day after NVIDIA. If you want exposure to this theme without underwriting a single counterparty, this is the vehicle. If you want it at a sane entry, you probably want it after August 27, not before.

Cerebras is a venture position that happens to have a ticker. Beta above five, no profits, 68 times sales, one customer, and an accounting quirk that will manufacture optical misses through 2031. There is a genuine technology edge here and the AWS channel is real. But a stock that travels from $386 to $161 to $252 in ninety days is not an investment, it is a position you size in basis points and defend with rules you wrote before you entered.

The broader test arrives in the next three weeks. NVIDIA on August 26, Marvell on August 27, Broadcom shortly after, Oracle in the second week of September, and Marvell's investor day on October 6. That sequence will tell you more about whether the buildout is being financed or merely announced than any backlog figure will.

Fair winds and following seas.

---

Disclaimer: This post is for informational and educational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any security. Prices, market capitalizations, valuation multiples, and analyst targets are approximate and as of mid-August 2026, and will change. Price-to-sales figures are derived estimates, not reported metrics. Backlog attribution to specific customers reflects analyst estimates and press reporting rather than company disclosure in every case, and should be treated as an estimate. Forward-looking guidance is management's, not the author's. All three securities discussed are volatile, and one is a recent IPO with limited trading history and a reported beta above five. Verify all figures against primary filings before acting. Past performance does not guarantee future results. Consult a licensed financial professional regarding your specific situation.

 SharkWater Trading  •  Income Strategies • Options • Private Markets

RVI: The Fund Where Nobody Knows What It Is Worth, And That Is Exactly Why You Sell Puts

August 17, 2026

Bottom Line Up Front

Robinhood Ventures Fund I (NYSE: RVI) went public at $25 in March, ran to $77.39 in May, fell all the way back to $24.10 in late July, and now trades near $28. The last audited net asset value the fund published was $24.05 per share, dated March 31. That number is four and a half months old.

Implied volatility sits near 87 percent against 30 day realized volatility of roughly 61 percent. When a market is charging you 87 percent vol on a fund whose true value updates once a quarter and holds a large slug of money market funds, the cleanest way to play it is not to buy the shares. It is to sell out of the money cash secured puts at or below the last known NAV and get paid to wait for a price that actually makes sense.

First, Almost Everything You Have Read About RVI's Holdings Is Wrong

Search "RVI holdings" and you will find confident pages telling you the fund owns Anthropic, xAI, Perplexity, Anduril, Scale AI, Figma, Notion, Discord, Chime, Brex, Plaid, and Mercury, with a $725 million NAV. Other pages will tell you the NAV is $300 million. One popular page insists RVI does not own SpaceX. Another insists SpaceX is the largest position at 22 percent of the fund.

None of that survives contact with the audited annual report. RVI's fiscal year ends March 31, and the Form N-CSR filed with the SEC lays out the entire schedule of investments, line by line, with cost basis and fair value. Nine private companies. Not twenty. Not one of the exotic names above.

It gets worse. Pull up RVI on some of the major quote sites and the fundamentals block still shows a market cap of $63 million, a 734 percent dividend yield, an industry classification of "Real Estate Operations," and a next earnings date in 2022. That is leftover data from Retail Value Inc., the shopping center REIT that used to trade under this ticker. The vendor never fully flushed it.

Rule of the boat: on a fund this new, with a recycled ticker and a quarterly valuation cycle, you go to EDGAR or the fund's own newsroom. Nowhere else. If your data source cannot tell you whether RVI is a venture fund or a strip mall landlord, it cannot tell you what it is worth.

What RVI Actually Owns

Here is the audited portfolio as of March 31, 2026, straight out of the schedule of investments, plus everything the fund has announced since.

Position	Type	Fair Value / Cost	Status
Databricks (Series K + L)	Preferred	$81.7M	Marked up 26.7% from entry
OpenAI	Class A Common	$75.0M	Added April 17
Revolut	Ordinary	$50.2M	At cost
Mercor	Series C Preferred	$50.0M	At cost
Whatnot	Preferred	$30.0M	Added Aug 5, $20B round
Airwallex	Series G Preferred	$25.0M	At cost
Boom Supersonic	Series B-1 Preferred	$25.0M	At cost
Oura	Series E Preferred	$25.0M	At cost
Ramp (common + preferred)	Both	$25.0M	At cost
Canva	Class A Common	$25.0M	Added June 24
ElevenLabs	Series D-1 Preferred	$20.0M	At cost
Stripe (+ June follow-on)	Class B Common	$14.6M plus	Follow-on June 29
SpaceX	Public common	$6.8M	IPO allocation, June
Money market funds	Level 1	$347.1M at 3/31	53% of net assets

Net assets were $655.3 million across 27,247,215 shares, which is where the $24.05 NAV comes from. The single most important line in that table is the last one. On March 31, more than half of this "venture fund" was sitting in a government money market fund yielding about 3.6 percent.

Since then the fund has deployed roughly $137 million into OpenAI, SpaceX, Canva, and Whatnot, plus an undisclosed Stripe follow-on. Back of the envelope, that still leaves somewhere in the neighborhood of $180 million to $210 million in cash equivalents, or roughly $6.50 to $7.75 per share. Call it a quarter to a third of NAV in T-bill proxies.

That matters enormously for a put seller. It means a meaningful chunk of what you would be buying on assignment is not a Level 3 guess about a private company. It is cash.

The NAV Problem, Which Is Really The Whole Trade

Here is the structural quirk that drives everything. RVI marks its portfolio and calculates NAV once per business quarter. The private positions are Level 3 fair value, determined by the adviser using precedent transactions, funding rounds, and secondary prints. The share price, meanwhile, trades every second the NYSE is open.

Think of it like a boat sitting on a mooring in fog. The NAV is a photograph somebody takes of the boat once every three months. The stock price is a crowd on shore shouting guesses about where the boat is right now. Sometimes the crowd is right. In May, the crowd decided the boat was worth $77 when the last photograph said $24.

The crowd was wrong. It usually is at extremes, in both directions.

The premium round trip on this thing has been violent:

Date	Price	vs Last Published NAV
March 6, IPO	$25.00 issue, opened near $22	Discount on day one
March 31	$26.54	+10% to $24.05 NAV
May 13, peak	$77.39	Roughly +220%
July 29, trough	$24.10	Roughly flat to NAV
Mid August	~$28.20	About +17%

A 68 percent drawdown from peak to trough in eleven weeks, on a fund holding money market funds and a dozen private marks. Nothing in the portfolio moved 68 percent. Only the story did.

Why The Options Are Rich

Metric	Reading
Implied volatility (30 day)	~86.6%
Historical volatility (30 day)	~61.3%
IV rank / IV percentile	~33% / 26%
IV high / low, trailing year	120.6% (June 9) / 69.6% (Aug 6)
Expected move, 10 DTE	+/- 11.8%, range $24.88 to $31.56
Put / call open interest ratio	1.21
Total open interest	~11,400 contracts

Roughly 25 volatility points of spread between what the market is charging and what the stock has actually delivered over the last month. That gap is your edge, and it exists for a reason worth naming out loud: the market genuinely does not know what this fund is worth between quarterly marks, so it prices uncertainty into every contract.

Note also that IV rank is only 33 percent. Vol is elevated in absolute terms but it is not stretched relative to this fund's own short history. It has been much higher. That argues for selling premium in size you can defend, not backing up the truck.

The Trade: Selling Out Of The Money Puts Below NAV

The setup writes itself. You have a fund trading at a 17 percent premium to a stale NAV, with roughly a quarter of that NAV sitting in cash equivalents, an 87 percent implied vol, and a demonstrated willingness to trade all the way back down to the low twenties. You do not want to buy $28 shares. You want to get paid for agreeing to buy $20 to $22 shares.

That is the entire thesis. You are selling insurance against a price you would happily pay anyway.

Strike Selection Logic

Anchor your strikes to NAV, not to the current share price. Three reference points:

• $24.05 is the last published NAV. Anything above that strike and you are agreeing to pay a premium to stated value.
• $24.10 is the July low. The market has already tested and defended roughly this level once.
• $21.00 is the 52 week low and represents about a 13 percent discount to the last NAV. At that level you are buying private venture marks at a real discount with cash backing part of it.

The ladder below uses model derived premiums at the observed volatility surface. These are theoretical values, not live quotes. Pull the actual chain before you place anything. On a name with 11,400 total open interest, the bid ask spread will take a real bite.

Expiry	Strike	Est. Premium	Breakeven	vs $24.05 NAV	Return on Cash
Sep 18 (32d)	$25.00	~$1.50	$23.50	-2.3%	6.1%
Sep 18 (32d)	$22.50	~$0.85	$21.65	-10.0%	3.8%
Sep 18 (32d)	$20.00	~$0.45	$19.55	-18.7%	2.3%

The sweet spot in my read is the October and November $22.50 strikes. They sit below the last published NAV, below the July low, collect a genuinely meaningful premium, and give you enough calendar to survive one full quarterly NAV print. The September $20 line is the sleep at night trade: small premium, but your breakeven is nearly 19 percent under stated NAV, and roughly a third of what you would own at that basis is money market funds.

The Catalyst Nobody Is Marking On Their Calendar

The March 31 shareholder update was published on May 29, about sixty days after quarter end. Which means the June 30 quarterly NAV update should land in the back half of August, along with the corresponding portfolio schedule filing. That is a defined, dateable event that will replace a four month old number with a fresh one.

This is the closest thing a closed end fund has to an earnings print, and unlike an earnings print, almost nobody is watching for it. If you are selling premium here, know that this date is inside your September and October contracts. That is a feature if you are short strikes well below NAV and a problem if you are short strikes near the money.

Bull Case

The fund is still roughly a quarter to a third in cash, which means the adviser has real dry powder to deploy into new rounds at current marks. The Databricks position is already marked 27 percent above entry, and Whatnot came in through a priced $545 million Series G at a $20 billion valuation, which is a hard valuation event rather than a model guess. There is no carried interest, which is a genuine structural advantage over a standard two and twenty venture fund. And if the OpenAI, Databricks, or Revolut positions get repriced upward at the June 30 mark, the NAV anchor moves up under the share price and the current 17 percent premium suddenly looks a lot more modest.

Bear Case

Three things worry me, in order.

One, the overhang. Robinhood Markets owned 52.18 percent of the fund as of March 31, and the fund has filed to register 14,217,271 of those shares for resale. That is more than half the share count sitting in a registered resale shelf above a thinly traded security. If that supply starts hitting the tape, the premium does not just compress, it inverts.

Two, the fee step up. The management fee is 2.00 percent of net assets, waived to 1.00 percent for the first six months after the IPO. That waiver runs out around early September. The all in expense ratio for the stub year already ran about 2.7 percent. On a fund holding a third in money market funds, you are paying venture fees on Treasury bills.

Three, Level 3 marks can gap. Every private position in this book is valued by the adviser using unobservable inputs. A down round at any one of Mercor, Boom, Oura, or ElevenLabs shows up as a step function in NAV, not a gradual drift. And RVII just listed on August 13 at $25 with roughly 80 Y Combinator names, which gives the same retail dollar somewhere else to go.

The SharkWater Take

I do not want to own RVI at $28. Paying a 17 percent premium to a NAV that is itself a quarterly estimate, in order to pay 2 percent a year on a portfolio that is one third cash, is not an edge. It is a subscription fee for a story.

But I very much want to own it at $20, and I am happy to be paid 5 to 11 percent on my collateral for saying so out loud. That is the entire trade. Sell the puts well below the last published NAV, size them so that assignment is a good day rather than a margin call, and let the 25 point vol spread work.

If you get assigned, you own a basket of Databricks, OpenAI, Stripe, Revolut, Canva, and a pile of T-bills at a real discount to stated value, and you flip straight into the wheel by selling calls into the next premium spike. If you do not get assigned, you keep the money and reload. Both outcomes are acceptable, which is the only kind of trade worth putting on.

Execution Notes

• Use limit orders, always. Average daily option volume is under 700 contracts across the whole chain. Market orders here are donations.
• Cash secured, not margin. A fund with a 68 percent peak to trough drawdown in eleven weeks is not where you want leverage on short puts.
• Size for full assignment. One contract equals 100 shares. At the $22.50 strike, that is $2,250 of real obligation per contract. Only sell what you can take delivery of without flinching.
• Close at 50 to 60 percent of max profit rather than riding to expiration. On a name with this much headline risk, the last 40 percent of the premium is the worst paid part of the trade.
• There is no dividend to cushion assignment. The fund has stated it does not anticipate being a predictable distributor. Your entire return on an assigned position is price plus whatever calls you write against it.
• Watch the quarterly update, not the ticker. Set an alert for the June 30 NAV publication and the corresponding portfolio filing. That number is the only thing that resets your anchor.

Fair winds and following seas.

---

Disclaimer: This post is for informational and educational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any security. Option premiums shown are model derived estimates, not live market quotes, and will differ from actual executable prices. Selling puts carries the obligation to purchase shares at the strike price and can result in losses substantially greater than the premium collected. RVI is a non diversified closed end fund holding illiquid Level 3 private securities whose fair values are estimates and may differ materially from realizable value. NAV figures cited are as of March 31, 2026 and are stale. Portfolio and financial data sourced from the fund's SEC filings and company announcements. Verify all figures independently before trading. Past performance does not guarantee future results. Do your own diligence and consult a licensed financial professional regarding your specific situation.

🦈 Novel Trading Ideas for the AI Titans: OpenAI vs. Anthropic Exposure, Relative NAV, and Proxy Trading Strategies

A slightly different way to trade.

The foundation model landscape in 2026 has fundamentally reshaped the definition of mega-cap tech. With OpenAI commanding an $852 billion post-money valuation and Anthropic filing for an IPO that targets an astronomical $965 billion, these private entities are structurally larger than almost every publicly traded pure-play software company on earth. For traders and portfolio managers, the challenge isn't recognizing the macro trend—it's determining how to efficiently trade private-market behemoths using public-market instruments.  

Here is a breakdown of the relative valuation landscape and the capital-preserving strategies required to trade the proxy spread.

The Tale of the Tape: Relative NAV and Growth Velocity

We are currently witnessing multiple expansions that completely defy traditional SaaS metrics.

 OpenAI: Following its unprecedented $122 billion funding round in March 2026, OpenAI's implied NAV sits at $852 billion. With an officially confirmed annualized revenue run-rate crossing $24 billion, the market is pricing the company at roughly a 35x to 39x revenue multiple.  

 Anthropic: The growth velocity of Anthropic has been staggering. After exiting 2025 at a $9 billion run-rate, Anthropic hit a $47 billion annualized run-rate by mid-May 2026. Valued at $380 billion in February, secondary markets and their Fall 2026 IPO filings point to a valuation approaching $965 billion. Despite the higher gross valuation, Anthropic's explosive revenue acceleration brings its forward multiple into a comparable orbit relative to OpenAI.  

The Public Proxy Ecosystem

Because pure-play access remains locked behind private secondary markets or impending IPO lockups, public equity exposure requires navigating the massive tech conglomerates. It is a tangled web of overlapping investments, compute agreements, and circular spending:

 Amazon (AMZN): Originally Anthropic's primary backer with an $8 billion stake, Amazon dramatically expanded its foundation model footprint by committing $50 billion to OpenAI's March 2026 round.  

 Microsoft (MSFT): The legacy proxy for OpenAI is now actively hedging its bets. Microsoft recently committed up to $5 billion to Anthropic, coupled with a $30 billion Azure compute deal.  

 Alphabet (GOOGL): Committed roughly $3 billion to Anthropic alongside a massive TPU supply agreement.  

 Nvidia (NVDA): The ultimate picks-and-shovels hardware play, investing $30 billion into OpenAI and $10 billion into Anthropic to ensure its GPUs remain the industry standard.

 SoftBank (SFTBY): A heavily leveraged proxy, holding over $71 billion in total OpenAI commitments.

Equity and Option Trading Strategies

Direct directional bets on AI via these public proxies can subject portfolios to heavy beta drawdowns if the broader market contracts. A sound, capital-preserving approach requires structuring trades that isolate the AI exposure or strategically monetize the inherent volatility.

1. Relative Value Pairs Trading

Rather than taking a naked long position in a tech giant, a market-neutral pairs trade can isolate the relative success of these foundation models while dampening broader market noise.

 The Trade: Long AMZN / Short GOOGL.

 The Rationale: Amazon has effectively cornered equity exposure to both OpenAI and Anthropic while driving massive AWS compute consumption across both ecosystems. Alphabet, while partnered with Anthropic, is simultaneously forced to defend its core search monopoly from both entities. This pairs trade capitalizes on Amazon's dominant compute-and-equity moat while neutralizing general tech beta.

2. Proxy Collars for Downside Protection

For those holding legacy long equity positions in Microsoft or Nvidia but wary of multiple compression or sudden AI capital expenditure fatigue, an options collar provides structured downside protection while maintaining exposure to the underlying growth.

 The Trade: Sell an out-of-the-money (OTM) Covered Call, and use the collected premium to finance the purchase of an OTM Put.

 The Rationale: This defines the exact risk/reward parameters. If the hardware or compute providers experience a rapid drawdown, the put rigidly caps the loss. The sold call caps extreme upside, efficiently financing the protection and maintaining a hedged posture.

3. Automated SPX Premium Collection

The S&P 500 has effectively become an AI proxy index. The companies providing compute, hardware, and capital to Anthropic and OpenAI command a massive weighting in the SPX. Instead of attempting to pick the winner between the foundation models, a highly robust strategy focuses on the systemic volatility their capital wars create.

 The Trade: Deploying automated SPX option rulesets (e.g., selling systematic short strangles or iron condors on the SPX index) while maintaining heavy cash positions.

 The Rationale: AI headlines—whether it is OpenAI's next funding round or Anthropic's IPO pricing—inject structural implied volatility into the broader index. By keeping capital secure in cash and deploying automated SPX premium collection strategies, traders can consistently harvest the inflated premiums generated by the AI hype cycle. You don't need to predict whether Anthropic or OpenAI wins the model war; you simply collect the premium generated by the market's uncertainty over the outcome.