Falling chip prices surprise the market, Nvidia introduces a platform to make quantum acceleration easier, and the U.S. Army takes another step forward on its key AI-powered node

Sign up
Work at CSET

We’re hiring! If you’re interested in joining our team, check out the positions in the “Job Openings” section below or consult our careers page.

Worth Knowing

As Chip Prices Fall, Semiconductor Companies Cut Their Orders: After two years of significant semiconductor shortages, prices of chips — especially commercial graphics processing units — have fallen considerably over the last two months. The reasons for the drop are likely manifold, but observers have pointed to two drivers in particular: falling demand in the face of high inflation and the recent cryptocurrency crash. The latter, especially, seems to have led to a glut in GPU supplies, with cryptocurrency miners selling off their hardware en masse as mining (a GPU-intensive task) became much less profitable. Whatever its causes, the news is welcome both for consumers — who saw the prices of cars and home electronics skyrocket for lack of essential chips — and AI developers, who rely on GPUs (whether their own or as part of a cloud) to train their models. But the sudden falloff in demand has thrown the world’s biggest semiconductor designers and manufacturers for a bit of a loop. Nvidia, AMD and Apple are all reportedly looking to cut their upcoming orders with chipmaking giant TSMC in response. It remains unclear what, if any, impact the easing chip shortage will have on proposed U.S. domestic chipmaking subsidies at the heart of the tech competitiveness bill working its way through Congress. While any new U.S.-based chip foundries wouldn’t come online for several years, the chip shortage had added urgency to the legislative effort. Without it, the bill — already facing hurdles due to disagreements among members of Congress over what does and doesn’t belong in it — could lose whatever momentum it had left.
Nvidia Tries to Bridge the Gap Between Classical and Quantum Computing: On Tuesday, Nvidia announced its Quantum Optimized Device Architecture (QODA), an open computing platform that aims to make hybrid quantum-classical computing more accessible for developers. While quantum computing will offer significant performance gains compared to classical computing in some important applications (including AI), pure quantum computers are not yet far enough along to have a broad practical impact. Nvidia says that embracing a hybrid approach, where quantum computers work together with classical, high-performance computers, is one way to take advantage of quantum computing’s benefits now. The company describes QODA as a “software bridge” that will allow developers to build quantum applications — whether by building completely quantum applications or by adding quantum computing to select, intensive tasks in existing applications — in an open, interoperable environment using either a hybrid quantum-classical system or emulated quantum processors running on Nvidia GPUs. In the same way that CUDA helped popularize GPU-accelerated computing, the company says it hopes QODA will help make quantum acceleration mainstream.  
Machine Learning’s Impact on the Large Hadron Collider: The European Center for Nuclear Research fired up the Large Hadron Collider last week for its third run. While machine learning played a role in the two previous runs, the LHC has undergone extensive maintenance and upgrades over the last three years that CERN researchers say will let AI and ML have an even bigger impact. As in previous runs, ML’s most important role will likely be in the realm of data filtering and processing; because proton collisions in the LHC produce much more data than can be recorded with current methods, LHC researchers use a “trigger” system to select out the most relevant data — approximately 100,000 events per second — for further analysis. Upgrades to the trigger’s sensor — new FPGAs (a powerful and flexible type of processor) allow CERN’s researchers to use filtering algorithms while preserving ultra-low-latency at the point of detection — and major improvements to CERN’s computing resources mean researchers can collect and analyze much more data than in previous runs. But researchers are also increasingly optimistic that unsupervised machine learning will be able to help them discover unexpected, previously unidentified phenomena. The current run is set to last until 2026, after which the LHC will be shut down for another period of maintenance and upgrades.
Government Updates

Cutting Off Chinese Chipmaking Supplies — U.S. Officials Lobby the Dutch: U.S. officials have asked the Dutch government to restrict exports to China of an important type of semiconductor manufacturing equipment, Bloomberg reported last week. Netherlands-based ASML Holding is the world’s largest manufacturer of photolithographic machines, a critical tool in modern semiconductor manufacturing. A push by U.S. officials during the Trump administration helped convince Dutch authorities to block exports to China of ASML’s crown jewel: its extreme ultraviolet lithography machines. EUV machines — which only ASML can make — are used in manufacturing the current generations of cutting-edge chips, and without them, Chinese chipmakers have to rely on an older type of “deep ultraviolet” lithography machine. But those older DUV machines can still produce chips as advanced as the 7nm node — more than capable of running advanced AI applications. In its final report last year, the National Security Commission on Artificial Intelligence recommended export controls on both EUV and DUV machines, but that would require the Dutch and Japanese governments to get on board. Bloomberg noted that a similar lobbying effort is underway with Tokyo, as Japan’s Nikon is another vital supplier of DUV machines. But the push could face an uphill battle — in a recent interview, the Dutch prime minister pushed back on efforts to “close off” his country’s trade relationship with China.

Army Awards Contracts to Prototype Key AI-Powered System: The Army announced that it had awarded $36 million each to Palantir Technologies and Raytheon Technologies to build prototypes of the Tactical Intelligence Targeting Access Node (TITAN), set to be essential to the military’s Joint All Domain Command and Control (JADC2) strategy. The strategy, which was approved last year by Defense Secretary Lloyd Austin, is the DOD’s ambitious plan to connect and coordinate all of its services’ sensors, rapidly process data — with the help of AI and machine learning — and accelerate decision-making. The Army sees TITAN as one of its key contributions to that effort — a mobile ground station, TITAN will collect data from ground, air and space sensors and use AI and ML to process it and quickly provide it to shooters in the field. The Army expects the prototyping phase will last 14 months, after which it will award a contract to a single vendor.    

Nuclear Regulator Wants Comments on Its Draft AI Plan: Earlier this month, the Nuclear Regulatory Commission — the U.S. agency responsible for regulating the use of nuclear energy and radioactive materials — opened up its draft Artificial Intelligence Strategic Plan for public comment. The 40-page plan is meant to cover the fiscal years 2023–2027, during which the NRC says it expects industry interest in using AI tools to grow. While researchers have already explored using AI to control plasma in a fusion reactor, the potential applications discussed in the NRC’s report are decidedly more behind-the-scenes. The report lays out five strategic goals meant to guide the NRC as it prepares to understand and regulate AI: to set up a “robust and flexible AI regulatory framework;” to build out an organizational capacity for processing AI applications; to work with industry, government, and academic partners to stay abreast of AI-related plans; to build up an “AI-proficient” workforce inside the NRC; and to explore potential use cases of AI to increase understanding. Comments on the draft plan will be accepted until August 19, and the NRC plans to hold a webinar on August 3 to discuss the plan and receive comments.  

In Translation
CSET’s translations of significant foreign language documents on AI

PRC Ministry of Foreign Affairs Budget: Ministry of Foreign Affairs 2022 Budget. This document is an original translation of the 2022 budget of China’s Ministry of Foreign Affairs, which manages China’s diplomatic efforts and its embassies and consulates overseas.

If you have a foreign-language document related to security and emerging technologies that you’d like translated into English, CSET may be able to help! Click here for details.

Job Openings

We’re hiring! Please apply or share the roles below with candidates in your network:
  • Research Fellow — Standards and Testing: This Research Fellow will focus on standards, testing, evaluation, safety and national security issues associated with AI systems. To do this, they will examine how the limitations, risk, and society and security impacts of AI can be understood and managed. Apply by August 15.
  • Research Fellow — AI Applications: This Research Fellow will focus on helping decision makers evaluate and translate new and emerging technologies, particularly in the field of AI, into novel capabilities by separating real trends and strategic opportunities from technological hope and hype. Apply by August 15.
  • Research Fellow — CyberAI: This Research Fellow will focus on exploring topics at the intersection of AI and cyber. AI techniques, specifically machine learning, offer an opportunity to uplift cyber defenses, but they also may create new threats to cyber infrastructure. This Research Fellow will examine potential machine learning (ML) applications to identify associated challenges and make relevant policy recommendations. A key to success will be the ability to explain the implications of these emergent capabilities to a non-technical audience. Apply by August 15.
Please bookmark our careers page to stay up to date on all active job postings. You can also subscribe to receive job announcements by updating your subscription preferences in the footer of this email.

What’s New at CSET

REPORTSPUBLICATIONSIN THE NEWS
What We’re Reading

Report: Cultivating Friendly Forces: The Chinese Communist Party’s Influence Operations in the Xinjiang Diaspora, Lin Li and James Leibold, Australian Strategic Policy Institute (July 2022)

Article: The Next Wolf Warriors: China Readies New Generation of Tough Diplomats, Richard McGregor and Neil Thomas, Nikkei Asia (July 2022)

Upcoming Events


What else is going on? Suggest stories, documents to translate & upcoming events here.
Schedule Update

policy.ai will be taking some time off for a summer vacation, but don’t worry — we’ll be back on August 11 with all the latest in AI, emerging tech and security policy!