Meta CEO Mark Zuckerberg has agreed to remain in California for eight years under a deal that links sales of new nonvoting shares to his residence, according to a regulatory filing released Wednesday. The arrangement follows his warning to directors on September 9 that he was considering leaving over Proposition 40, a ballot measure proposing a one-time 5 percent tax on billionaires' assets held as of January 2026. For Meta, the pact matters because it keeps its founder near management and Meta Superintelligence Labs while funding AI data centers without diluting voting control.

Zuckerberg Ties Share Sales to Eight-Year Stay in California

Board deal after warning about Proposition 40

Zuckerberg told board members at a September 9 dinner that he was weighing departure and planned to list his Palo Alto home for sale four days later, the 109-page filing stated. The proposal at issue could face lawsuits even if voters approve it next month, yet it would apply to stock and other assets held in the state. Like many wealthy holders, Zuckerberg limits current tax payments by borrowing against shares while taking a $1 annual salary. He had purchased a home in Florida this past February, as reported by The Wall Street Journal at the time.

Within days of that dinner, a special committee reached an agreement with Zuckerberg that was finalized a week ago and disclosed on Wednesday. Meta plans to create a new class of shares with no voting power, distributed automatically to all existing holders in proportion to current ownership. Under the Founder Alignment Agreement, Zuckerberg may sell 20 percent of those new shares to cover potential tax liabilities. The remaining 80 percent becomes available in portions over eight years only while he stays in California, with early sales requiring disposal of voting shares except in material unforeseen circumstances.

The committee included attorney Robert Kimmitt, venture capitalist Charlie Songhurst, and DoorDash CEO Tony Xu. It cited difficulties during the COVID-19 pandemic when Zuckerberg could not be physically present with managers and sought to avoid a repeat. The filing defines residence as wherever Meta keeps its headquarters, leaving room for a joint relocation. That detail sits against departures by other California firms to Texas, Nevada and other states, plus reported exits by Google cofounders Sergey Brin and Larry Page, while pioneering AI work remains concentrated in Silicon Valley around Menlo Park.

What the residence pact means for AI business

For companies building or buying AI systems, the structure shows how founder proximity is treated as an operating asset during heavy infrastructure spending. Meta wants Zuckerberg close to leadership and the unit behind recent Muse AI models while it raises funds for data centers. Smaller firms cannot replicate a new share class, but they can formalize where key technical leaders work during critical build phases. Larger groups with distributed teams face the same trade-off between executive mobility and day-to-day oversight of model and infrastructure programs.

The limits deserve attention before drawing broader conclusions about tax-driven relocation. Polling suggests Proposition 40 will fail after about $129 million was spent campaigning against it, so the tax may never take effect. The agreement does not prove that staying in California preserves AI talent or project speed, and it does not remove litigation risk around the measure. Buyers and partners should ask what triggers the unforeseen-circumstances exception, how sale tranches are scheduled, and whether headquarters location could change without breaking the terms.

The marker to watch is the vote next month and Meta's subsequent share actions. Rejection of the measure would reduce the immediate tax pressure while leaving the eight-year residence and staged-sale schedule in place. Approval followed by court challenges would test whether the funding plan and leadership arrangement hold under uncertainty. Either outcome will show how durable location-linked founder agreements prove when AI investment cycles demand both capital and constant management presence.