Due Diligence Capital Notes

When a company files an 8-K — a form the SEC requires for any material event — announcing its annual meeting voting results, most investors scroll past it in a matter of seconds. Nothing blew up. Everyone got re-elected. Moving on.

I think that's the wrong instinct, at least when it comes to Core Scientific (CORZ). The company's May 15, 2026 Annual Meeting 8-K contains no revenue figures, no hash rate disclosures, no update on the HPC contracts that the market is desperately trying to size up. What it does contain is a vote tally — and that tally, read carefully, tells a story about shareholder unease that I think is worth sitting with.

Here's the short version: roughly one-third of voting shareholders pushed back on executive pay, two board members drew meaningful opposition approaching one-in-four votes, and yet the CEO sailed through with barely 1% withheld. That's not a random distribution. It's a signal.

What an Annual Meeting Proxy Vote Actually Measures

Before dissecting the numbers, let me explain what we're looking at. Once a year, public companies hold an Annual Meeting where shareholders vote on a slate of governance items — usually director elections, executive pay packages, and auditor ratification. The results get disclosed in an 8-K filing, typically within four business days of the meeting.

The votes on director elections and executive compensation are rarely close. Most years, board nominees get re-elected with 90-95% support and pay packages pass comfortably. When the numbers deviate sharply from that baseline, it's worth asking why.

Three specific mechanisms are at play in this filing:

Director elections: Shareholders vote "For" or "Withhold" on each nominee. Withholding is the only real protest mechanism available to a dissatisfied shareholder in an uncontested election — there's no "Against" option. A withhold rate above 20% is widely considered a meaningful rebuke in governance circles.
Say-on-pay: A non-binding advisory vote where shareholders signal approval or disapproval of named executive officer (NEO) compensation — the CEO, CFO, and other senior leaders named in the proxy statement. "Non-binding" means the board isn't legally required to change anything, but persistent large-scale opposition tends to attract activist attention and proxy adviser pressure.
Auditor ratification: Shareholders vote to keep or reject the external accounting firm. This one almost always passes overwhelmingly, because switching auditors mid-stream is expensive and disruptive.

Understanding these three mechanisms is essential to reading what happened at Core Scientific's May 12, 2026 Annual Meeting.

Breaking Down the Vote Tallies

Let me walk through each proposal with the actual numbers, because the contrast between them is the whole story.

1. Director Elections: A Tale of Two Outcomes

All five nominees were elected to serve through the 2027 Annual Meeting. On the surface, that reads as a clean sweep. But the vote distribution underneath is anything but uniform.

Adam Sullivan (CEO): 217,650,785 shares voted For, 2,469,826 withheld. That's a withheld rate of roughly 1.1%. By any governance standard, shareholders gave the CEO their near-unanimous blessing as a board member.
Elizabeth Crain: 215,074,279 For, 5,046,332 withheld — about a 2.3% withheld rate. Also unremarkable.
Yadin Rozov: 213,054,408 For, 7,066,203 withheld — roughly 3.2%. Still comfortably within normal range.
Jeff Booth: 168,580,185 For, 51,540,426 withheld — approximately 23.4% withheld.
Eric Weiss: 168,947,230 For, 51,173,381 withheld — approximately 23.2% withheld.

Booth and Weiss each crossed the 20% threshold that governance watchdogs treat as a formal warning sign. At major institutional investors like State Street and Vanguard, a nominee crossing 20% withhold triggers internal flags and can lead to voting against re-election the following year if the board doesn't respond visibly.

Why Booth and Weiss specifically? The 8-K gives us no explanation — it's a pure vote tally filing, not a narrative document. But the symmetry is striking: both drew almost identical withheld vote totals, suggesting coordinated institutional dissatisfaction rather than random individual objections. Booth is well-known as an author and Bitcoin macro thinker; Weiss has a background in finance and digital assets. Whether the objection is to their committee assignments, their compensation committee decisions, or their broader alignment with the HPC/AI strategic pivot is something we'd need the proxy advisory firm reports to fully understand.

2. Say-on-Pay: A Meaningful Protest

The say-on-pay vote — the non-binding shareholder advisory on whether named executive officer compensation is appropriate — passed with 138,785,477 votes For and 79,166,695 Against, with 2,168,439 abstentions. That translates to roughly 36% opposition among voting shareholders.

To put that number in context: a say-on-pay opposition rate above 30% is the threshold at which proxy advisers like ISS (Institutional Shareholder Services) and Glass Lewis typically recommend against the compensation committee members in the following year's election. An opposition rate approaching 40% starts to appear in annual governance failure studies published by major asset managers.

The filing confirms the technical outcome in language required by SEC Item 5.07: "The Company's stockholders approved, on a non-binding, advisory basis, the compensation of the Company's named executive officers, as disclosed in the 2026 Proxy Statement." Approved, yes. Enthusiastically endorsed? Clearly not.

Core Scientific is in the middle of an expensive strategic transformation — pivoting from pure Bitcoin mining to high-performance computing (HPC) and AI infrastructure, a capital-intensive build that requires long-term executive retention. That context makes the pay friction more consequential. The last thing you want when committing hundreds of millions in capital expenditures over a multi-year runway is leadership uncertainty driven by compensation disputes.

3. The Broker Non-Vote Bloc

One number in the filing deserves more attention than it typically gets: 44,778,216 broker non-votes on the director election and compensation proposals.

A broker non-vote occurs when a brokerage firm holds shares on behalf of a client but hasn't received voting instructions from that client. Brokers can vote uninstructed shares on "routine" proposals (like auditor ratification), but they cannot vote them on "non-routine" proposals like director elections or say-on-pay. Those shares get counted toward quorum but don't affect the outcome of contested matters.

Why does this matter? That 44.8 million share bloc represents a substantial pool of passive or uninstructed institutional ownership. In any future contested vote — a special meeting, a contested director election, an activist campaign — these shares could decisively swing the outcome depending on how their beneficial owners respond to a direct solicitation. Right now, they're quiet. They won't stay that way forever if governance friction compounds.

4. KPMG: The One Clean Result

Contrast everything above with the auditor ratification: 262,833,296 votes For KPMG LLP, 313,161 Against. That's a 99.88% approval rate, essentially unanimous. The filing notes: "The Company's stockholders ratified the selection by the Audit Committee of the Company's Board of Directors of KPMG LLP as the independent registered public accounting firm of the Company for its fiscal year ending December 31, 2026."

This is the governance equivalent of a clean EKG. Shareholders have no problem with who is auditing the books. Their concerns are elsewhere — specifically, in who is setting executive pay and who is sitting on the board.

What This Means for an HPC/AI Buildout Story

Core Scientific's investment thesis right now is not primarily a Bitcoin mining story. The company has been repositioning as an HPC/AI infrastructure provider — leasing its data center capacity and power infrastructure to hyperscalers and AI compute customers. That's a fundamentally different business than proof-of-work mining: longer-term contracts, lower-volatility revenue, but also massive upfront capital requirements and multi-year execution risk.

In that context, governance isn't a sideshow. Governance is part of the thesis.

If the board is divided — or if a meaningful slice of institutional shareholders believe the board isn't properly aligned with long-term value creation — that friction will eventually show up in management decisions, capital allocation debates, and retention of key executives needed to execute HPC contracts. CEO Adam Sullivan's near-unanimous director support is genuinely reassuring. But the concentrated skepticism toward two specific board members, combined with the pay opposition, suggests that some institutional investors are watching this evolution very carefully and are not yet satisfied with what they're seeing.

What Could Break This Thesis

No governance-friction analysis is complete without naming the specific failure modes that would turn an amber signal into a red one.

Compensation restructuring disrupts retention. If the board — responding to the 36% say-on-pay opposition — restructures executive pay packages mid-cycle, it risks dislodging the leadership team at exactly the wrong moment. Losing a CFO or CTO during active HPC contract negotiations would be operationally damaging and likely market-moving.
Booth and Weiss draw higher opposition in 2027. Proxy advisers have long memories. If the compensation committee (which Booth and/or Weiss may sit on) doesn't visibly respond to the 23% withheld signal, ISS or Glass Lewis could recommend withholding support in 2027 — this time with explicit institutional following, potentially pushing withheld rates above 40% and triggering a formal governance review.
The 44.8 million broker non-vote bloc gets activated. Right now, those shares are quiet. If an activist fund builds a position and directly solicits those beneficial owners before a future meeting, they become a swing vote of significant size. A 44.8 million share swing in a contested election or shareholder proposal could alter outcomes meaningfully.
Operational reality disappoints. This filing contains zero information about hash rate, HPC contract signings, revenue trajectory, or capital expenditure progress. The governance friction is visible; the operational story is still largely opaque from this data point alone. If upcoming earnings releases or filings reveal HPC contract delays or cost overruns, the governance skepticism visible in these votes could crystallize into something more aggressive.

Looking Forward

The May 12, 2026 Annual Meeting didn't produce any dramatic outcome. All five nominees are seated. KPMG is in for another year. The pay package passed. But the distance between "technically approved" and "genuinely endorsed" is wider at Core Scientific right now than most headlines suggest.

What I'm watching for in the next several quarters is whether the board demonstrates that it heard the signal. Visible changes to compensation structure, more transparent disclosure around HPC contract economics, or a public acknowledgment of the pay-vote outcome in the next proxy statement would all suggest the board is managing this friction proactively. Silence, by contrast, tends to invite escalation.

Core Scientific is executing on one of the more credible HPC/AI infrastructure pivots in the digital assets space. The underlying strategic logic is sound. But strategy and governance are not separate variables — they compound each other in both directions. A leadership team with genuine shareholder alignment behind it executes with more latitude. One operating under persistent governance friction finds that friction tends to show up at the worst possible moments.

The vote tallies from a routine annual meeting 8-K told me something real. I'd rather know this now than wait for it to surface in a proxy fight.