The BIG picture on AI chips

There are multiple bottle necks for AI Chips. Without chip dominance no country or company can lead the AI tech charge. What should we being doing to stay ahead of China and still have precautions so AI doesn’t turn into terminator.

Chips are no longer just a tech supply issue. They are now a national power issue. Whoever controls advanced chips controls the speed of frontier AI, military AI, robotics, and future economic productivity.

Right now, the landscape has three major forces:

The U.S. is trying to preserve its AI advantage.

The U.S. has used export controls to limit China’s access to advanced AI chips and chipmaking tools. That includes restrictions around high-end GPUs, chipmaking equipment, and access by Chinese firms outside mainland China.

China is pushing hard for self-reliance.

China’s goal is to reduce dependence on Nvidia, AMD, ASML, TSMC, and other Western-aligned chokepoints. China recently returned to the top of the TOP500 supercomputer list with a domestically designed chip system, although experts said that system is not optimized for modern AI workloads.

The Netherlands is joining the U.S.-led Pax Silica initiative, which is meant to coordinate AI and semiconductor supply chains among allied nations. This is especially important because the Netherlands is home to ASML, the dominant supplier of advanced lithography machines needed for leading-edge chip production. The U.S. and Netherlands agree on restricting the most advanced tools to China, but there are still tensions over how far those restrictions should go.

This has the makings of a pseudo cold war which could turn into a hot war in taiwan.

With governance being our specialty here at the AI Steward lets spend some time on what I believe should be done at the national level to both protect us AND keep us ahead.

1)Treat compute as another area of AI Governance. Most AI governance policies focus on model use, data privacy, bias, and procurement. That is good, but add compute.

2) Create geopolitical risk algorithms that have human in the loop decision making. Just like you are using the Janis Pointe dash board, the people making decisions should have non biased and public information regarding risk assessment of choices as a whole. The future of governance is ging to be human with AI intelligence and this is just the start.

3)Avoid single vendor dependence. For god sakes, I understand that TSMC is the best chip maker in the world however whatever these trillion dollar companies need to do to match them, do it. The government should provide allowances to incentivise this so WW3 doesn’t occur.

4)Require compute and energy impact disclosures. We need more ways of getting AI information out to the public. Let us all understand why we need more data centers and how we can create win wins by building them. At minimum, have a well executed narrative (which has been abysmal for data centers)

AI chips are becoming the oil, railroads, and nuclear supply chain of the AI era. They determine who can build the most powerful systems, who can deploy at scale, and who becomes dependent on whom. More communication, less dependence will equal a WIN for America on all fronts.