Concentration risk is any single exposure or group of exposures with the potential to produce losses large enough (relative to capital, total assets, or overall risk level) to threaten a financial institution’s health or ability to maintain its core operations. (from https://ncua.gov/regulation-supervision/letters-credit-union...)
80% Of OpenAI And Anthropic’s Enterprise Revenues Come From 1% Of Its Customers, Which Skew Heavily Toward AI Startups Subsidized By Venture Capital
Anthropic and OpenAI Are Dependent On Artificial Revenue Driven By Unprofitable Venture-Backed AI Startups For Billions Of Dollars Of Revenue
I agree, the headline lacks the nuance of what Zitron includes in the footnotes. What is your estimate of concentration risk of AI use in VC-backed startups? Do you think there is more enterprise use that is not captured and it's more like 40% or 20%. I am not trying to be argumentative, I am trying to get a sense of other viable alternatives.
I'm not sure how it would change it, the 1% firms number would probably get even more concentrated if you included Microsoft and trading firms like Jane Street (assuming they opted out), but the spend weighted share of VC backed companies would probably go down. However, large VC backed stealth companies may be more likely to opt out. And lots of Microsoft's spend might be on products and support/devops servicing demand from VC backed stuff.
From a recent Dwarkesh interview with Dylan Patel it sounded like Jane Street might be double digits of Anthropic's revenue, but I'm not sure where they sourced that.
Microsoft and Jane Street are paying for AI out of profits, so that cash flow looks more sustainable than investor dollars flowing to frontier labs via VC-backed startups. But it's a good point that some of Microsoft's spend may also be the result of VC investment in startups. I guess the fundamental questions is if either OpenAI or Anthropic can find a sustainable model. Probably, but not in their current configuration.
The "we help you fire your people" messaging has been a catastrophe but there have clearly been breakthroughs in code generation with more to be discovered.
I suspect AI becomes more like the steam engine, the railroads, or electricity, transforming society but leaving it fundamentally recognizable. Reasonable men may differ.
"There is plenty of international talent that can and should be leveraged in the appropriate places and the proper ways. There are many instances where you’re going to want to hire a leading expert in a technical field. But most jobs, the qualifications are a core technical and subject competency combined with diligence and a willingness to learn. [..]
We need to encourage a corporate mindset that believes in the importance of building up the American talent pool to tackle not only the challenge in front of them but the array of challenges that the future will bring."
> But most jobs, the qualifications are a core technical and subject competency combined with diligence and a willingness to learn. [..]
> We need to encourage a corporate mindset that believes in the importance of building up the American talent pool to tackle not only the challenge in front of them but the array of challenges that the future will bring.
It's tough to square this noble thought with what's actually happening in corporate America. The culture of investing in training and mentoring to foster employee growth has been eroding for the past 40 years (if not more), to where it only seems to exist in limited capacity in legacy institutions and is completely absent in new (tech) firms. How does one convince business leaders to bring it back?
> The culture of investing in training and mentoring to foster employee growth has been eroding for the past 40 years (if not more), to where it only seems to exist in limited capacity in legacy institutions and is completely absent in new (tech) firms. How does one convince business leaders to bring it back?
With engineers in the big tech companies mostly only staying for maybe 1-2 years I can understand employers not being very interested in training and mentoring to foster employee growth.
40 years ago engineers frequently stayed at companies like DEC, IBM, and HP for 10 to 20+ years.
> With engineers in the big tech companies mostly only staying for maybe 1-2 years I can understand employers not being very interested in training and mentoring to foster employee growth.
This is an effect, not a cause. People started voting with their feet when companies stopped investing in them.
> How does one convince business leaders to bring it back?
There are precisely 1.0 ways to do that: make all of the alternatives more expensive.
That's it. Nothing else will move the needle.
I don't believe it is possible: every policy one can imagine, short of establishing a GULAG system to compel compliance, has powerful opposing interests, including workers themselves. It would require the sort of comprehensive and sustained focus that a nation aspiring to nearly suicidal liberalism can not deliver.
Every system I can imagine as a real unintended consequence that is worse than what we have. The ideal should move to needle back - but there is no way to make that happen that isn't worse than where we are now.
When nobody is going to work for the same company for decades what is the point? This is both management's fault for not valuing the people they have (it is very common for cost of living raises to not match inflation or much less increased value as people get better), layoffs; but also workers for looking for a new job (only partially because of the above) when they have one.
I think in the ideal world 30-60% of the people who retire from your company should have started there just out of school. Much less than that and you don't get the value of experience built up over time. Much more and you miss out of good ideas others are doing. However there are many ways the world will never be ideal...
That is easy to say. You can put it on a political mailing "vote for me and I'll"... However details matter. I've yet to see any detailed proposal that isn't obviously worse than where we are now.
> We need to encourage a corporate mindset that believes in the importance of building up the American talent pool to tackle not only the challenge in front of them but the array of challenges that the future will bring."
Ergo, laws will be required to "encourage" the corporate paperclip maximizer apparatus. When did asking nicely ever work? "Do not fall into the trap of anthropomorphising the lawnmower."
Of course, but there’s a wide gulf between slamming into a wall at 80mph and gently rolling to a stop. The latter requires foresight and action, however.
I don't think the famously young-skewed tech industry is at any risk of slamming into a wall. If Google finds in 2035 it's running out of good seniors in their 30s because they haven't been mentoring college grads well enough, there'll be plenty of experts in their 50s and 60s to pull back in from management or FIRE or age discrimination.
Thank you for the kind words. I remember everything I’ve ever read, it’s simply how fast I can aggregate links and threads based on keywords from HN search and my internal knowledge base/graph.
Objective data, evidence, and facts are important as a shared foundation to discuss upon.
Gupta, a distinguished computer scientist from Harvard, is a serial entrepreneur and currently a partner of Y Combinator. Before joining YC, he was the co-founder and CTO of Reveries Labs. The controversy that he landed in stemmed from his comment on JD Vance's post."If an American corporation needs workers, it should hire and train Americans," Vice President JD Vance wrote on X, reacting to the Donald Trump administration's new H-1B visa fee.
"in a win for the woke left DEI is back," Ankit Gupta commented, suggesting that if Americans have to be hired because they are Americans and not because of their skill, then it is DEI (Diversity, Equity and Inlcusion) only.
As Gupta defended the H-1B hiring by Silicon Valley, he asked one commentator: 'what elements of H-1B immigrant culture do you not like? is it the willingness to work long hours? Or perhaps the education focus of their children?"
Gupta has locked down his tweets so the his remarks and the replies are no longer available.
Key takeaways: Founders should treat meetings as crucibles for getting facts, perspectives, and potential courses of action on the table for a full and frank discussion that results in a working consensus. Some advice bootstrapping founders managing the dynamics of a team meeting with four to sixteen people. A two- or three-person conversation will also benefit, but there is a phase transition that seems to occur at four to five people up to about 16 for dealing with significant decisions.
+ Address issues promptly but prepare: sleeping on it and putting your thoughts in writing can often make for a more productive conversation.
+ Listen and check understanding: ask clarifying questions and accurately summarize what you heard to confirm shared understanding.
+ Stay calm and respectful, and avoid defensive responses. Take a break to maintain equanimity.
+ Keep your statements brief and on point. Don’t interrupt, and don’t repeat yourself unless asked to clarify.
+ Don’t suggest solutions until you have a rough consensus on a problem or need.
+ Don’t select a solution until you have evaluated at least two alternatives to the status quo.
+ It’s OK to disagree on the situation or a course of action in any one conversation or sequence of conversations: commit to finding a workable solution.
I can accept the author's premise that LLMs double coding productivity today.
Semiconductors following Moore's Law increased their density, which is a reasonable metric for effectiveness, by 40% a year for several decades. Leaving aside projections of AGI in two years, it's still a substantial impact if we can figure out how to increase coding productivity by 40% a year for two or three decades. I don't think that gets us to AGI or the Singularity, but it's similar to the impact of the steam engine, steel, or electricity.
These are "normal technologies" that were transformational, and that may be the path we are on with LLMs.
tl;dr: Google DeepMind's AlphaEvolve improved the speed of Substrate's computational lithography stack by 680% and reduced compute costs by 97%, enabling the simulation and printing of the most demanding layers in advanced semiconductor devices.
Water filtration / sewage management in a city
Electric illumination transformed everyday life, improved working conditions, etc..
I think you are ignoring technology as infrastructure having a transformative impact on life expectancy and quality of life.
reply