RIOT - Ultrathink: The Electricity Paradox Revisited

A Buffett/Munger meditation on Riot Platforms, power as infrastructure, and the difference between having an asset and monetizing it

The Core Question: Has Anything Changed?

Three weeks ago, we rejected Riot Platforms as a value investment. The business failed every quality test. The moat was nonexistent. The financials were catastrophic. The AI/HPC pivot was 99% aspiration and 1% execution.

Today, Riot's stock is up 21% to $16.89. Meanwhile, Bitcoin has fallen 18% to ~$72K, the company sold 3,778 BTC to fund operations, and its Chief Data Center Officer -- the single most important executive for the pivot thesis -- walked out the door, forfeiting $18.7 million in unvested shares. The fundamentals deteriorated while the stock rallied.

This is the kind of divergence that makes thoughtful investors uncomfortable, and it should. When price and value move in opposite directions, one of two things is happening: either the market knows something you do not, or the market is wrong. Understanding which requires intellectual honesty of the highest order.

Moat Meditation: Power is Necessary but Not Sufficient

Munger taught us to think about competitive advantage through the lens of durability. Not "do they have an advantage today?" but "will they still have it in ten years?"

Riot's 1.7 gigawatts of approved, interconnected power capacity in Texas is genuinely scarce. Getting a new gigawatt permitted and connected takes 5-7 years in most jurisdictions. This is not a trivial asset. In the AI buildout, power has become what silicon was in the 1960s -- the binding constraint on growth.

The AMD lease provides the first tangible evidence that this power can be monetized at attractive economics: $12.4 million per megawatt of contract value, generating roughly 2.5x more gross profit than Bitcoin mining. If you could wave a magic wand and contract all 1,700 MW at similar terms, Riot would be worth $20-40 per share. The market knows this, which is why the stock trades where it does.

But here is where Munger's skepticism becomes essential: having a scarce asset is not the same as having a moat, and having a moat is not the same as being a good investment.

Consider the parallel with oil reserves. Saudi Arabia sits on the world's most valuable oil reserves. But that does not make Saudi Aramco a great investment for minority shareholders -- because the entity controlling the asset has different incentives than outside investors. In Riot's case, the "entity" is a management team that has: (a) destroyed $3.5 billion in cumulative free cash flow, (b) diluted shareholders by 117% in 2.5 years, (c) lost its key data center executive after 10 months, and (d) is currently selling its Bitcoin treasury to cover operating costs.

Owning 1.7 GW of power is necessary for the thesis to work. But the question Buffett would ask is: "Do I trust this management team to convert that asset into shareholder value?" The evidence says no.

The CDO Departure: A Signal Worth Hearing

Buffett has always said that when management insiders are selling, investors should pay attention. The corollary is that when key executives depart and forfeit millions in compensation, the signal is even louder.

Jonathan Gibbs left after ten months. He forfeited $18.7 million in restricted shares. The company has not explained why. The filing calls it "mutual."

In my experience, "mutual" departures that involve forfeiting eight-figure compensation packages are rarely mutual. Something went wrong. Possible explanations include:

1. Strategic disagreement: Gibbs may have wanted to move faster on HPC conversion, or slower, or in a different direction than the board
2. Pipeline disappointment: The Corsicana hyperscaler discussions may not be as advanced as the market hopes
3. Cultural clash: Converting a Bitcoin mining company into a data center company requires fundamentally different operational DNA
4. Better opportunity: In a market where data center executives are in extreme demand, Gibbs may have simply received a better offer

We cannot know which explanation is correct. But we can observe that Riot now has zero dedicated data center operating leadership at the C-suite level during the most critical phase of its strategic pivot. This is not a minor organizational matter -- it is the difference between a company executing a pivot and a company talking about executing a pivot.

Risk Inversion: What Must Go Right?

Munger's inversion is particularly illuminating here. Instead of asking what could go wrong, let us ask what MUST go right for RIOT at $16.89 to be a good investment:

1. BTC must recover above $90K to make mining economics viable again. Currently below the $77K industry breakeven
2. A replacement CDO must be hired quickly and be of sufficient caliber to negotiate hyperscaler deals
3. At least 200-500 MW of additional HPC leases must be signed within 12-18 months
4. Share dilution must slow significantly -- the 117% dilution in 2.5 years must stop
5. No further BTC treasury liquidation beyond normal treasury management
6. Competitors must not lock up all available hyperscaler demand before Riot executes

That is six things that must go right simultaneously. The probability of any single one is perhaps 50-70%. The probability of all six is much lower -- perhaps 10-20%. This is the math that separates speculation from investment.

The Patient Investor's Calculus

Here is the honest assessment, stripped of both narrative enthusiasm and value investing purism:

RIOT is not a value investment. It never will be one. It is a leveraged option on two things: Bitcoin's price and AI power demand. Both of these are real secular forces, but neither creates a durable competitive advantage for Riot specifically.

However, the AMD lease changed the analysis from "pure speculation" to "speculation with evidence." A $311 million contract with a Fortune 500 company is not nothing. It proves the concept works. The question is execution scale, and at the right price, even uncertain execution is worth buying.

That price is not $16.89. At $16.89 and $6.4 billion market cap, you are paying $3.8 million per megawatt for a portfolio that is 98.5% uncontracted, managed by a team without a data center operating chief, mining a commodity below breakeven cost, and funding operations through treasury liquidation.

The right price is closer to $7-10 -- near book value, where you are essentially paying for the BTC treasury, the cash, and the AMD lease, with the remaining 1,675 MW of power capacity as free optionality. With 84% annualized volatility and the stock having traded at $6.19 within the past year, this entry is not hypothetical -- it is a realistic patience trade.

A Note on the Aschenbrenner Position

Leopold Aschenbrenner's Situational Awareness LP holds RIOT at 1.8% of portfolio. This is worth understanding but not overweighting. First, 1.8% is a small position -- it is not a high-conviction bet. Second, Aschenbrenner's thesis is not about Riot specifically but about power infrastructure broadly. He holds multiple miners and infrastructure plays. Third, Aschenbrenner's expertise is in AI timelines, not in evaluating management quality or data center execution capability. His insight that "power is the bottleneck" is probably correct. His implicit bet that Riot specifically will capture that value is much less certain.

Conclusion: The Waiting Game

The hardest thing in investing is distinguishing between a good asset in bad hands and a bad investment. Riot Platforms has a genuinely valuable asset -- 1.7 GW of approved Texas power capacity. But that asset is controlled by a management team with a track record of value destruction, run through a corporate structure that has diluted shareholders by 117% in 2.5 years, and is currently losing key executives during its most critical strategic transition.

The correct response is not to reject the opportunity entirely -- the asset is too valuable for that. Nor is it to buy at current prices, which embed significant execution optimism that the evidence does not support. The correct response is to wait, with discipline and patience, for the market to offer this asset at a price that compensates for all of these risks.

That price is $7-10 per share. At that level, you own the BTC treasury, the cash, and the AMD lease for approximately book value -- and you receive 1,675 MW of uncontracted power capacity as a free option on the AI infrastructure thesis. That is an asymmetric bet worth making.

Above $10, you are paying for management execution that has not been demonstrated. And paying for undemonstrated execution is the very definition of speculation.

"The stock market is a device for transferring money from the impatient to the patient." -- Warren Buffett. In the case of RIOT, the patient investor waits for Bitcoin weakness to provide what the business itself cannot: a margin of safety.