Breaking

Vinod Khosla Calls for U.S. Tax Code Overhaul as AI Threatens Labor Income

September 21, 2026 at 18:04 EDT

Policy & Regulation
Sales & Business

Vinod Khosla is urging the U.S. to redesign its tax code for an AI economy in which machines take on a large share of economically valuable work, arguing that policy should protect displaced workers rather than preserve existing jobs. In a Sept. 14 essay, the Khosla Ventures founder and early OpenAI backer framed AI-driven job disruption as a looming test for capitalism, public services and the political legitimacy of rapid AI adoption.

September 2026 · U.S. policy

Khosla's AI Tax Reset

The proposal treats AI labor disruption as a fiscal problem: shift more tax burden from workers toward capital and possible AI usage, then return proceeds to people hit by displacement.

64%

Share of all human labor in Khosla's 80% by 80% estimate

$400B

Estimated annual receipts from capital gains and tax-break changes

20%

Conditional token tax on AI compute and robotic substitution revenue

$15T

U.S. GDP component Khosla says came from labor in 2024

Where the first $400B would come from

$300B

Capital gains changes

$100B

Tax-break changes

Takeaway: the larger pillar is projected to come from ending preferential capital-gains treatment; the smaller pillar comes from closing selected business and inheritance-related preferences.

Policy sequence

Tax capital

Treat gains like ordinary income and remove selected preferences.

→

Fund transitions

Send proceeds first to workers hit by AI displacement.

→

Add AI levy if needed

Apply compute or robotic-substitution taxes only if disruption is significant.

Policy appeal

Supporters see the plan as a way to keep AI deployment politically viable by insuring workers and expanding low-cost services.

Open questions

Critics question the displacement timeline, the link between task exposure and job loss, and how a compute tax would be measured.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.