AI Policy Wiki
Dashboard

Center for Strategic and International Studies (CSIS)

high confidence · updated 2026-07-23

Washington think tank; home of the Wadhwani AI Center and Gregory Allen's export-controls analysis, the most-cited open-source policy commentary on the US-China AI compute race.

The Center for Strategic and International Studies (CSIS) is a bipartisan nonprofit policy research institution in Washington, DC, founded in 1962. It produces analysis across defense, foreign policy, economics, and emerging technology. Its dedicated AI policy unit is the Wadhwani AI Center, directed by Gregory C. Allen, whose writing on AI export controls and the US-China compute race is widely cited.

FieldValue
TypeBipartisan nonprofit policy research (think tank)
HeadquartersWashington, DC
Founded1962
President / CEOJohn J. Hamre (long-tenured)
AI focus unitWadhwani AI Center
Known forDefense, foreign policy, and technology analysis; export-controls writing; China AI/compute coverage

Overview

CSIS produces bipartisan, policy-facing analysis across defense, foreign policy, economics, and emerging technology. On AI, its focus is US national-security competitiveness: compute, export controls, alliance dynamics, and the US-China technology race.

Wadhwani AI Center

The Wadhwani AI Center at CSIS is the institution's dedicated AI policy unit. It was founded in 2022 with backing from Romesh Wadhwani's foundation. Its coverage spans export controls and compute governance; US-China AI competition and DeepSeek-era narrative debates; defense AI (procurement, autonomy, and Department of War posture); and allied coordination with the UK, EU, Japan, Korea, and India.

The center is directed by Gregory C. Allen, a former founding director of strategy and policy at the DoD Joint AI Center (JAIC, now Chief Digital and Artificial Intelligence Office (CDAO)). Allen is among the most-cited US commentators on AI export controls. His work reframed the post-DeepSeek narrative from "controls failed" to "controls have design flaws and enforcement gaps, but the core premise remains sound" (see CSIS — DeepSeek, Huawei, Export Controls, and the Future of the U.S.-China AI Race (Allen, March 2025)).

Activities

Two AI-relevant publications anchor the center's output. DeepSeek, Huawei, Export Controls, and the Future of the U.S.-China AI Race (Allen, March 2025) is a primary-voice analytic defense of the post-DeepSeek US export-control regime. The CSET/CSIS US-China national-security AI brief is a joint brief by Arnold, Horowitz, and Allen covering Task Force Lima, the AI RCC, CDAO, and PLA AI procurement patterns.

A CSIS-led wargame in which House Homeland Security and China select committee lawmakers weighed responses to Chinese AI-enabled cyber threats to critical infrastructure became public on July 23, 2026 (Source: insideaipolicy.com).

Positions

CSIS has repeatedly argued that post-October-2022 BIS controls have materially constrained Chinese compute, but that BIS under-resourcing, smuggling networks, and design loopholes such as the A800 must be closed. It treats compute as the primary leverage surface and aligns closely with CSET and RAND on compute governance; Allen is frequently cited alongside Heim and Patel.

A central CSIS intervention in the 2025 debate is that DeepSeek did not falsify export controls: DeepSeek's genuine innovations coexist with continued Chinese compute constraints. Beyond AI, CSIS has a broader tradition of arguing that the US defense industrial base is under-resourced relative to the China challenge.

Funding and governance

CSIS is funded by a mix of corporate sponsors, foundations, and foreign governments. The Wadhwani AI Center is anchored by the Wadhwani Foundation; CSIS also receives funding from defense primes and major tech companies. Critics note the mix of corporate and government funding and the think tank's historically hawkish defense positions; defenders cite bipartisan staffing and methodological transparency.

Relationships

Sources