This quick take draws on research presented at the 62nd Eastern Academy of Management: "Comparative Case Research on CEO Compensation: NVIDIA vs. AMD vs. INTEL" (Gagne et al., 2025), co-authored with Owen Gagne. Here is what the data made me think — beyond what belongs in a conference paper.
All three companies claimed a pay-for-performance philosophy. Only two of them actually lived it. NVIDIA's reliance on pure PSUs — with no RSU safety net and genuinely stretching performance thresholds — is what performance-contingent compensation is supposed to look like. Jensen Huang's equity grew because NVIDIA's results were exceptional. That's the design working.
Critically, Huang is the only one of the three CEOs who received no cash bonus even in a year of exceptional company performance. In 2023 — a year in which Nvidia posted 239% TSR and its stock price surged — Huang received no cash bonus. NVIDIA's Compensation Committee sets threshold, base, and stretch performance levels for variable cash, and those targets were not met on the specific metrics governing bonus eligibility that year. This is the discipline that distinguishes a genuine pay-for-performance structure: equity rewards follow market outcomes, but cash bonuses follow pre-established operational targets — and if no target is met, no bonus is paid, regardless of how well the stock performed.
Su's balanced structure at AMD similarly aligns pay to outcomes rather than tenure. Her stock options pay nothing unless AMD's stock appreciates above the exercise price. Both structures demonstrate that a well-designed pay-for-performance framework does not require complexity; it requires consequence.
Gelsinger's exit has been analyzed largely through a strategy-and-execution lens. But the compensation structure around his tenure deserves equal scrutiny. The core design problem was not in the metrics themselves — revenue growth, cash flow, and TSR are all appropriate for a turnaround — it was that the structure prioritized market competitiveness over strategic accountability.
Intel's Compensation Committee benchmarked Gelsinger's package against market peers. The result was a $140M new-hire package front-loaded with equity that reflected what the market paid for top semiconductor leadership — not what Intel's actual turnaround performance warranted. When a company designs pay to be 'market competitive' without building in sufficiently aggressive performance gating, it effectively decouples compensation from results. The 80/20 PSU/RSU split looked balanced on paper, but the RSU component guaranteed equity payouts regardless of performance — reducing the urgency the moment called for.
Shareholders voted against the 2021 executive compensation program. The 1,171:1 CEO-to-median-worker-pay ratio that year was a visible symptom of the misalignment. When Gelsinger's relative TSR PSUs — 368,965 shares from the new-hire grant — were forfeited in March 2024 due to below-threshold performance, it confirmed what the design had obscured: the compensation structure had rewarded the promise of a turnaround, not its delivery.
Intel's current leadership and its Compensation Committee now face a defining design challenge. A credible reset would include a new-hire package that is milestone-gated rather than front-loaded, heavier PSU weighting with genuinely stretch thresholds, meaningful clawback provisions tied to strategic execution, and a reduced RSU component that does not cushion the CEO from the consequences of strategic failure. The 2026 proxy season will be an important signal of whether the board has learned from the Gelsinger chapter.
Across all three compensation packages analyzed — spanning four fiscal years — non-financial metrics are essentially absent from equity incentive design. TSR, operating income, revenue growth, EPS. That is the universe. For companies that are now central to national AI strategy, global supply chain resilience, and the energy consumption profile of data centers worldwide, this is a notable omission.
The business case is real: institutional investors are increasingly scoring governance quality on ESG integration in compensation. Emerging tax policy frameworks — already operational at the city level in Portland, increasingly discussed at the federal level — will likely create financial incentives for companies that demonstrably link executive pay to sustainability outcomes.
NVIDIA has an energy-efficient GPU architecture that could enable measurable carbon targets. AMD has workforce diversity gaps in technical roles it could incentivize closing. Intel has an explicit RISE initiative with 2030 targets already on record. None of them has yet made these targets count in the compensation system. That gap will narrow — the question is whether boards get ahead of it or respond to it.
The SEC's PVP disclosure rules, clawback requirements, and the growing momentum behind say-on-pay transparency are not temporary compliance burdens — they represent a structural shift in the governance environment that compensation committees must now design for, not around. The forfeiture of Gelsinger's TSR PSUs is exactly the kind of visible, disclosed consequence that regulators and shareholder advocates have been trying to create through these mechanisms. More of this is coming.
At the same time, AI itself is beginning to reshape how executive performance is evaluated. As semiconductor companies become the infrastructure layer of the AI economy, compensation structures will need to account for how executive decisions around model partnerships, data center investment, and chip architecture choices translate into durable competitive positioning — with time horizons that may exceed even the current three-to-four-year vesting windows that dominate equity design today.
The companies that treat compensation design as a forward-looking strategic tool — not a retrospective retention exercise — will be better positioned to attract the leadership profiles the chip industry demands in its next decade.