> Today, both the country’s new tech firms, like DeepSeek, and existing powerhouses, like Huawei, are increasingly keeping pace with their American counterparts.
This one sentence seems to be the extent of the "falling behind" from the headline?
The authors have some concrete policies they want to see adopted:
> launch a large-scale AI literacy initiative across the government
> invest billions of dollars in procurement over the next few years
> expand support for the National AI Research Resource
The article exists to convince someone with the power to direct those billions of dollars to direct them this way. Claiming that the US is falling behind is a popular trick to make that happen.
Presumably the holders of purse strings know that policy papers must not be taken at face value, but if they're unaware that the robots in Chinese factories are far from the AI-driven humanoids of popular imagination, or that the AI Plus Initiative is full of lofty goals with nothing concrete on how to achieve them, they might be fooled nonetheless.
The issue is, most of the decisionmakers on the Hill still have an image of China that is comparable to where it was in the 1990s or 2000s.
Most decisionmakers started their careers in the 1980s to 2000s and only worked within the bubble that is the Hill, and most of their assumptions are predicated on the experiences of an American who was either in or adjacent to the academic and cultural elite of the 1990s and 2000s.
Those people with domain experience have limited incentive to work as staffers or within think tanks because they do not hire broadly, they pay horribly, and domain expertise is only developed through practical experience, which takes a decade to develop.
That is not to say this isn't an issue in other countries (even Chinese and Korean policymakers have fallen into similar traps), but most other countries also try to build an independent and formalized civil and administrative service. The American system is much more hodgepodge and hiring is opaque (eg. The nebulous "federal resume"), meaning most people hired will have went to schools where career services provide training to join government jobs (eg. Top private schools along with public universities in the DMV).
The issue in the US is a coordination issue - we have the right mixture of human, financial, and intellectual capital, but it is not being coordinated.
Not at all. A big thrust of the article is about falling behind in AI adoption. See the first 3 paragraphs below the heading "Innovation and Adoption." Specifically:
>"Although the United States and China are very different and the latter’s approach has its limits, China is moving faster at scaling robots in society, and its AI Plus Initiative emphasizes achieving widespread industry-specific adoption by 2027. The government wants AI to essentially become a part of the country’s infrastructure by 2030. China is also investing in AGI, but Beijing’s emphasis is clearly on quickly scaling, integrating, and applying current and near-term AI capabilities."
China is facing a demographic cliff that is potentially catastrophic.
I remember Japan talking about replacing its similar demographic problems with robots.
Didn't happen. Now ai and robotics is apparently progressed... But I'm guessing this will be some grand vision in the CCP to save their country, while at the same time fulfill the CCPs great desire for a totally controlled and subservient workforce.
Much like the Cold war, there's a lot of scare that can be built into that. Which corporations can use to get a whole lot of sweet government and military money.
But almost everything that was held up as an existential threat to democracy in the USSR turned out to be overblown in the best case, an outright fraud or smokescreen frequently.
As we can see from the Ukraine invasion, corruption in the military and control structures follows these authoritarian regimes. China also has this problem.
China was functioning well under reduced Deng Xiaoping rulership, but Xi is a typical purge and control authoritarian, which implies bad things about China's long term economic health.
Between the authoritarianism, demographic cliff, and possibly a massive real estate/finance bomb, China will probably have to become expansionist.
But they have nuclear frenemies all on their borders: Japan (effectively), Russia, Pakistan, India. They can be blockaded from petroleum access with a single us carrier group, which will happen if they invade Taiwan and I don't think they can help themselves.
>They can be blockaded from petroleum access with a single us carrier group, which will happen if they invade Taiwan and I don't think they can help themselves.
I doubt it. This is just an armchair general's cope that's so faulty that I don't know where to start attacking it from.
Since Clinton's 1990s show of force in the Taiwan straits, China has built up a formidable navy, esp. their submarine forces. Within the straits, their AA/AD web, SOSUS, etc. guarantees they have freedom of action. Moving further, they have a strong submarine component that can seriously threaten any blockading forces.
They're also the world's sixth largest oil producer, and Naval War College [0] estimates suggest that they can stretch their emergency reserves to 8 years if they enforce, say, 45% rationing.
That's before you factor in that you'd be blockading up to 60% of the world's seaborne cargo volume, both from China, Japan, South Korea, etc. Unprecedented in the whole of human history.
Then there's their formidable 5th-gen. air force that can dominate South Korea, Japan, etc. easily, nuclear weapons that guarantee they won't lose any territory, and the massive economic whiplash the entire West will face as a result.
India might hate China, but it wants a multipolar world and won't help the West cut them down to size. So, they won't assist with the blockade.
The rest of Southeast Asia's economies are deeply interconnected with China, and they don't want to stir up their wrath, so they might condemn them, but won't even wave a stick at them.
I just hope the Pentagon's policymakers and war planners are less myopic. if you stumble in a confrontation with this overconfidence, and still lose Taiwan to China, Pax Americana is effectively over. The humiliation will be impossible to recover from.
A lot of that is predicated on the fact that a CS major in China (and other countries like India, Israel, the CEE) also studied CompArch, OS, and other "low level" disciplines which we in the US don't treat as CS anymore, which leads to a lack of understanding of how to integrate hardware with software.
The fact that DSP is a CSE major requirement abroad, but optional in much of the US aside from ECE programs (but even they have now gated DSP to those ECEs who want to specialize in EE) highlights this issue.
Can't reply so replying here:
> There are lots of young whippersnappers and “old timers” in the “west” who could easily do the Low level make it quick on small hardware stuff
Not to the same degree. The total number of CE graduates (from BS to PhD) is 19k per year in the US.
A large number of those were not introduced to table stakes CS classes like programming language design or theory of computation.
Conversely, for CS major, they are not introduced to intro circuits, digital logic design, DSP, comp arch, and in some cases even OS development because there was a pivot in how CS curricula for undergrads was designed over the past 10 years.
> in the context of adoption as opposed to frontier development.
For real world applications like military applications or dual use technology, frontier development is not relevant. It's important but it's not what wins wars or defines industries.
Being able to develop frontier models but being unable to productionize foundational models from scratch for sub-$2M like Deepseek did despite paying US level salaries highlights a major problem.
And this is the crux of the issue. The best engineers are those who recognize what is "good enough".
Americans who did their undergrad here over the past 10 years act more like "artists" who want to build to perfection irrespective of whether it actually meets tangible needs or is scalable.
> We aren’t actually engineers, we didn’t get to take classes in the engineering college, maybe we should have
Which is the crux of my argument.
CS is an engineering discipline, and some of the best CS undergrad programs in the US like Stanford, Cal, MIT, and UCLA make sure to enforce Engineering requirements for CS majors.
The shift of CS from being a department within a "College of Engineering" to being offered as a BA/BS in the "College of Arts and Sciences" sans engineering requirements is a recentish change from what I've seen.
> Incidentally a lot of AI movers are EEs, not even CSE or CEE.
Yep! Gotta love Information Theory and Optimization Theory. And a major reason I feel requiring a dual-use course like DSP for CS/CE majors is critical.
There are lots of young whippersnappers and “old timers” in the “west” who could easily do the Low level make it quick on small hardware stuff, the US companies just aren’t asking us to?
Computer science doesn’t have the EE pre-requisites to do DSP while computer engineering does. We aren’t actually engineers, we didn’t get to take classes in the engineering college, maybe we should have.
Incidentally a lot of AI movers are EEs, not even CSE or CEE.
ECE is a compromise track that limits how far one can progress on hard systems design problems. An ECE’s principal strength is that they tend to manage long term software projects better than EEs.
EEs lead in anything that is limited by physics because that is how the discipline is oriented. The bleeding edge of compute has been like this since the first microprocessors were created, at the very latest.
I am not disputing or arguing the the reasons for it. I was simply pointing out that the "falling behind" part in the article was more in the context of adoption as opposed to pure development.
The intention of the Sullivan Doctrine was to keep Chinese players 1-2 generations behind in AI/ML applications, as this would give breathing room for the US, because AI/ML is essentially an HPC problem that can be solved by throwing a ton of capital and compute.
The issue is, a large subset of American "AI" startups are founded by people who are ideologically driven by an almost religious fervor around unlocking AGI and superintelligence.
On the other hand, most Chinese startups in the space are highly application driven and practical in nature - they aren't chasing "AGI" or "Superintelligence" but building a monetizable product and outcompeting American players. (P.S. Immigrant founded startups in the US approach the problem in the same manner)
I've said this a ton of times on here, but most American MLEs are basically SKLearn wrapper monkeys with a penchant for bad philosophy. It's hard to find MLEs at scale in the US who understand both how to derive a Restricted Boltzmann Machine as well as tune and optimize the Linux Kernel to optimize Infiniband interconnects in a GPU cluster.
Most CS and CE majors in the US who graduated in the past 7-10 years think less like engineers (let's build shit that works, and then build it at scale) and more like liberal arts majors but wanted to learn enough coding to pass leetcode medium and get a job - I've had new grad SWEs who are alumni of MIT caliber schools ask me about how to become a VC or PM and how I did the SWE-PM-VC transition because "they don't want to code". I was gobsmacked.
The same mindset occurs abroad as well in China, India, Eastern Europe, Israel, etc but at least they force students to actually learn foundations.
And if you look at most of the teams who lead or develop either GPU architecture, high performance networking, RL research, Ensemble learning research, etc - most did their undergrad abroad in China, India, or the CEE but their PhD in the US. The pipeline and skill at the junior level in the US is almost nonexistent outside a handful of good programs that are oversubscribed.
When (picking a random T10 CS program) Cal and UIUC CS tenure track professors are starting to take up faculty positions in China and India's equivalent of those CS departments, that means you have a problem.
CS is a god damn engineering disciple. Engineering is predicated on bridging the gap between theoretical research and practical applications, but I do not see any backing for this kind of mindset in most American programs.
And yes, the work ethic in the US leaves much to be desired. If Polish, Czech, and Israeli engineers will be fine working 50-60 hour weeks during crunch time, asking you to work earlier hours in order to accommodate your private commitments after 3PM is not some form of egregious abuse.
The American tech industry has become lazy, the same way the American automotive industry became lazy in the 90s and 2000s. The lack of vision and the pettiness amongst management and the lack of motivation amongst ICs who are amongst the highest paid in the world is not conducive if we want to retain a domestic tech industry.
And unlike the automotive industry of ye olde days, the tech industry being a services industry can and has begun moving P/L and product roadmap responsibilities along with the execs who own said responsibilities abroad. If the HQ is in the US, but all the decisions are made abroad, are you really an American company?
A big part of the problem is management at American firms. They are rarely, if ever, run by engineers at the helm. If you put arts and business majors in charge, it's no surprise that outputs look like art and business projects. These leaders pick people just like them at all tiers. Those who do boring and honest engineering work are shunned, excluded from promotions and left out of the leadership circle. It's little wonder that all of the real engineers depart for greener pastures.
Fix leadership and you will fix American industry.
I think the problem lies with the American polity, values, and business environment, and not industry leadership per se. Smart new grads generally go where the money is, and for the last 20 years that has meant either finance or big data firms that may have no interest in real technical progress.
> Smart new grads generally go where the money is, and for the last 20 years that has meant either finance or big data firms
Software TC has outpaced high finance for almost 15 years now, especially for the kinds of candidates who had the option between the two.
I went to one of those universities where CS grads had the option between being a Quant at Citadel, an APM at Google, or an SWE working on an ML research team. Most CS students chose 2 and 3 because the hours worked were shorter than 1 and the hourly wage and TC was largely comparable.
> may have no interest in real technical progress.
Hard to make technical progress as (eg.) a cybersecurity company when most CS programs do not teach OS development beyond a cursory introduction to systems program, and in a lot of cases don't introduce computer architecture beyond basic MIPS.
The talent pipeline for a lot of subdisciplines of CS and CE has been shot domestically for the past 10-15 years when curricula were increasingly watered down.
> I went to one of those universities where CS grads had the option between being a Quant at Citadel, an APM at Google, or an SWE working on an ML research team
I find it interesting that you are trying to find fault in my position while describing the same phenomenon in even greater detail than I could have myself. All of the roles you list are broadly aligned with the finance or adtech industries, which simply do not employ people with the skills you desire in sufficient numbers. The talent you seek is following the money, just like your peers did.
The talent pipeline was shut down because adtech sucked most of the oxygen out of the room, and consolidation by hyperscalers finished off what was left. Big data firms simply don’t need the same headcounts to maintain their infrastructure - it is arguably the whole point of their business.
I have spent over a decade asking myself when the systemic cost of this would be realized. Better now than in another 10 years - all of the cohort that predates this will have aged out of the workforce by then.
> The talent pipeline was shut down because adtech sucked most of the oxygen out of the room, and consolidation by hyperscalers finished off what was left
I've been in this space as an IC, a Manager, and a VC and trust me when I say the education standards have been watered down in CS for 10 years now, that I no longer have a pipeline to train detection engineers, exploit developers, eBPF developers, and others out of college in the US.
Just take a look at the curriculum changes for the CSE major (course 6-3) at MIT in the 2025 [0] versus 2017-22 [1] versus pre-2017 [2] - there is a steady decrease in the amount of EE/CE content and an increased amount in math. Nothing wrong with increasing the math content, but reducing the ECE content in a CSE major is bad given how tightly coupled software is with hardware. We are now at a point where an entire generation of CSE majors in America do not know what a series or parallel circuit is.
And this trend has been happening at every program in the US over the past 10 years.
Management culture has issues, but in the tech industry, management has been technical in nature for a generation now.
I've funded startups in Israel and the US, and trust me when I say that the mindset of the average IC engineer in Israel versus the US is a night and day difference.
The Israeli IC will be extremely opinionated and will fight for their opinions, and if it makes sense from a business perspective, the strategy would change. But the Israeli IC when fighting these battles would also try to make a business case.
On the other hand, when I used to be a SWE, I almost never saw my peers try to fight for engineering positions while also leveraging arguments supporting the business. That's why I became a PM, but I noticed the same IC SWEs like the former overwhelmingly became PMs. And then a subset of those PMs become founders or VCs like I did.
I've found solutions and sales engineers to be the best management track individuals - technical enough to not be bullshitted by a SWE who really really loves this specific stack, but also business minded enough to drive outcomes that generate revenue.
But anyhow, the point is there is a mindset issue amongst Americans across the entire gamut of the American tech industry - especially amongst those who started their careers in the past 10 years.
This is the reason. I logged in to basically say the same thing. I used to be this way, and give opinions, but you cause problems over time that ends up with you getting disciplined in subtle ways or fired.
Yeah, it’s hilarious to be having this conversation about MLEs while attributing the bad outcomes to anything other than poorly designed reward functions, i.e. management. If an engineer burned millions on failed training runs because they did a shit job of creating a policy that maximized for the desired outcome, they’d get canned, but that’s just a Tuesday for your average MBA with VC backing.
It’s more nuanced than that. Americans on the west coast can express dissent; the means, however, are indirect and easily missed by managers who lack cross cultural competence. There’s also less motivation for a worker at a large successful American firm to express dissent in the first place. Employees at smaller firms speak up earlier.
The Israeli IC will be extremely opinionated and will fight for their opinions, and if it makes sense from a business perspective, the strategy would change. But the Israeli IC when fighting these battles would also try to make a business case.
That’s not because they have different engineering perspectives, that’s an Israeli cultural trait. Israeli’s tend to index more towards directness in their communication. That’s definitely not the case with someone from, say, India.
The top 8 are US companies, no 9 being TSMC. I'm sure you've come across people within the industry who are lazy but there are no doubt a lot who are not.
Rising market caps only reflects interest in public capital to flow into (public) tech companies.
And the composition of the 10 largest market cap tech companies hasn't really changed for over a decade - the only new entries are Nvidia, Broadcom, TSMC, and Tesla.
Sklearn hasn’t been relevant for at least 5 years now. Anyone doing anything serious with it is committing malpractice since there are lots of far better, faster, alternatives to it. In particular, Nvidia RAPIDs cuML.
This one sentence seems to be the extent of the "falling behind" from the headline?