The AI PhD Stipend Debate: Is $50,000 Too Much?

Episode 1 · Debate Erupts Over Whether $50K PhD Stipends Are Fair (2026-11-09, 2 posts)

A debate over PhD stipends questions whether $50K for AI PhD students is excessive, with Stanford's Anshul Kundaje pushing back that critics ignore living costs, while Toronto PhD students note their stipends were only $16-23K until recently.

• U Toronto AI PhD stipends were just $16k-23k until recently, top now ~$30k — Afinetheorem · 2026-11-09
• Stanford prof rebuts PhD stipend "injustice" claim: NYC/Bay costs ≠ Moscow/Mumbai — anshulkundaje · 2026-11-09

Episode 2 · Economist Argues PhD Students Are Costly Apprentices, Not Workers (2026-11-10, 6 posts)

On November 10, economist Afinetheorem posted a series of tweets pushing back on the popular claim that "grad students keep universities running and PhD students are workers," arguing it bears no resemblance to reality: PhD students are fundamentally students—and extremely expensive ones for the school—closer to costly apprentices than employees.

Confirmed

• He ran the cost math: PhD programs are small, generate no revenue, and carry high advising/training costs; at a school like Harvard, training one PhD student costs roughly $500,000.
• He shared internal emails from Kevin Bryan, a professor at the University of Toronto's Rotman School of Business, advocating for PhD stipends as corroboration.
• He noted that judged purely on their contribution "as labor," the net return to universities from admitting PhD students is negative—under pure employment logic, schools would admit none; no school would pay $100,000, training costs included, for a "PhD worker."
• He cautioned student organizers: universities treating PhD students as major cost centers is an objective fact.

Why it matters

• The debate directly affects the rhetorical foundation for PhD stipend and unionization movements: if PhD students are framed as "trainee apprentices" rather than employees, the logic of demands for higher pay needs reworking
• Afinetheorem isn't opposed to raising stipends, but argues the budget facts must be acknowledged before making the case—he frames it as an opinionated debate about the economics of academic labor, not simple side-picking

• Economist: Grad students aren't workers, they're costly students — Afinetheorem · 2026-11-10
• Professor: a Harvard PhD student costs the school ~$500k — Afinetheorem · 2026-11-10
• Professor: a PhD student costs a school like Harvard ~$500k — Afinetheorem · 2026-11-10
• AI researcher: universities would hire zero PhD students as workers, they cost more than they provide — Afinetheorem · 2026-11-10
• AI researcher on PhD stipend debate: raise stipends if you like, but acknowledge the budget facts — Afinetheorem · 2026-11-10
• Professor on PhD stipend debate: students are cost centers, not hired workers — Afinetheorem · 2026-11-10