Lorand Laskai

currently grants @coeff_giving. formerly @SenateBanking, NSC @WhiteHouse46, @CommerceGov

[Chris McGuire @ChrisRMcGuire]

While Nvidia beats sky-high expectations every quarter, China’s second-biggest AI chip designer’s Q3 revenue missed revenue expectations by a massive 44% due to capacity constraints and memory shortages. China’s AI chipmaking remains massively constrained due to export controls.

Interestingly, the last time Nvidia missed revenue expectations was August 2022–one month before US export controls on AI chips went into effect.

[Emily Feng 冯哲芸 @EmilyZFeng]

Trying something new - my op-Ed in the @washingtonpost is based off extensive reporting on Uyghurs in Syria, and Beijing’s chance to break a cycle of violence https://t.co/YQ3ByATmwP

[Lorand Laskai @lorandlaskai]

Treasury just proposed a very significant expansion of its outbound programs. A lot to unpack, but the following proposed expansions jump out on an initial read:
1️⃣ Broadly prohibits investment in developing AI/super intelligence, irrespective of end-user and amount of compute used in training
2️⃣ Prohibits investment in biotechnology, broadly, including development of pharmaceutical compounds.
3️⃣ Aligns with export controls in prohibiting investments in development or production of high-bandwidth memory.
https://t.co/6FWl8oFhbX

[Caleb Watney @calebwatney]

Seems bad that AI company leaders are currently more scared of Jensen than of Congress!

[Chris Murphy 🟧 @ChrisMurphyCT]

Read this. An advertisement for a new American industrial policy. China will eat our lunch without one.

[Tim Fist @fiiiiiist]

Big new IFP report from @GeorgiaCAdamson, @KonstantinPilz, and me on the US-China AI competition.

We spent the last 6 months digging into how Chinese AI companies’ access to compute and distillation of American models translate into the current US-China capabilities gap.

The result is a 23k-word research report, including an interactive analytical model that visualizes changes in the US AI lead under different policy interventions.

Our findings:

1. The US AI lead over China is currently only around 6 months.

2. The US could roughly double this lead, to over a year, using authorities already available to the federal government.

3. On chips: China’s access to AI chips will roughly double over the next year, with chips housed in data centers overseas accounting for roughly half of China’s 2027 compute.

4. On distillation: the current ~16x US training compute advantage would normally yield a 13-month lead. But we estimate adversarial distillation halves this to around 6.5 months.

5. Increasing the lead would have some obvious benefits:

- Give the US broadly a greater cyberdefensive advantage against Chinese threat actors.

- Give US AI companies breathing room to spend more R&D resources on making their systems safer and more secure, without fear of being overtaken by less safety-conscious Chinese competitors.

6. Provide the US government more leverage in negotiations with China about the future of AI.

7. The two most important interventions are:

- For BIS: Create a new end-use control to restrict China’s remote access to training clusters abroad — China’s largest and fastest-growing source of frontier-scale compute.

- For DOJ/FTC: Create a narrow safe harbor to allow US companies to coordinate on countering Chinese adversarial distillation.

8. Both interventions (and all the others we discuss in the report) are things the federal government could do today.

9. To estimate the effects of each policy intervention, we model the US lead as a function of its compute advantage over China and China’s ability to adversarially distill the capabilities of the US frontier.

10. The model makes many simplifying assumptions, but provides a useful way to assess and rank the impact of different policy interventions at a high level.

Read the full report here: https://t.co/7pC9ZKtikx

[Lorand Laskai @lorandlaskai]

The first rule of export controls is don't control items with foreign substitutes. China's new controls suggest MOFCOM didn't get the memo.

These controls will be a boon for boring and dredging firms, primarily in Europe, that have struggled to keep up with Chinese competitors.

[Kyle Chan @kyleichan]

China is rolling out even more export controls.

Just days after draft export controls on AI, biotech, and solar, China is now adding or updating export controls on dual-use items:
- Drones
- Construction machinery
- Dredging equipment

[Kyle Chan @kyleichan]

China: We believe in openness and cooperation on technology and development—unlike some countries.

China: We’re planning to place export controls on an additional 40 technologies, covering AI, biotech, and solar.

[Lorand Laskai @lorandlaskai]

@ChorzempaMartin Assuming the HFAC version becomes law, there is still a wide universe of tools China will continue buying from the US, JPN, and NLD, including dry ArF litho, which is perfectly suitable for ~90 percent of China’s current production needs.

[Lorand Laskai @lorandlaskai]

Beijing’s biggest coup in recent years is convincing western analysts that it’ll burn down the global economy if liberal democracies take actions that are in their self-interest.

I don’t think China will shut down the global SME industry in response to MATCH. Here’s why:

[Martin Chorzempa 马永哲 @ChorzempaMartin]

This is what Match act proponents are missing. If China is fully shut off from machines, they can and will shut off the flow of critical material to US/NL/JP equipment makers and then production grinds to a halt. No one will get them.

[Lorand Laskai @lorandlaskai]

Fair question! The Senate version is from April and obviously doesn’t reflect ongoing NDAA negotiations.

The version HFAC voted out of committee is a better representation of where things are heading - core restrictions on DUVi remain, but some of the ancillary controls will fall out.

[Lorand Laskai @lorandlaskai]

We’re thrilled to have you!

[Ruby Scanlon @rubyscanlon]

Some personal news!

I'm thrilled to be joining the new U.S. AI Policy team at Coefficient Giving (@coeff_giving)

Americans are increasingly concerned about the implications of powerful AI. Last month alone, AI’s salience among voters rose as much as it had over the entire previous year

Today, large bipartisan majorities back government action to reduce risks from AI, even when primed about risk from China. But translating public concern into effective policy will require strong institutions, deep expertise, and the capacity to grapple with the challenges ahead

This new team will help seed and strengthen—as @calebwatney has described—institutions wise and capable enough to help us navigate the transition to powerful AI

I couldn't be more excited about the work ahead!

[Lorand Laskai @lorandlaskai]

And ultimately, the US and its allies have the means to punch back if they need to. They can ratchet up export controls, impose sanctions, limit market access. Beijing does not have a monopoly on imposing economic pain.

[Lorand Laskai @lorandlaskai]

The fact that the US and allied actions are not without costs means we should be prudent in which action we take. I think @KhanSaifM make a strong case for why MATCH is worth the cost of upsetting Beijing. https://t.co/iiJ4XiwLa7