Hacker News

> Theoretically Apple can spend just as much. What are the outcomes though?

The GP was talking about Google specifically, and their outcomes on AI are nothing to scoff at. They had a rocky late start, but they seem to have gotten over that. Their models are now very much competitive with the startups. And it's not just that have more money to spend. They probably have more training data than anyone in the world, and they also have more infrastructure, more manpower, more of a global footprint than the startups.

The Innovator's Dilemma is an anecdotal, maybe a statistical relationship at best, but not a fundamental law of nature. When an established company has everything it should take to become a leader in a new industry in theory, and in practice their products are already on par with the industry leaders, you know at some point it becomes rational to think that maybe they might become a leader.

Sometimes I worry about the incentives for innovation in the US.

Step 1, find something to innovate on, sell the promise of it to investors. Step 2, build a prototype or worst case, build it for real and start generating income from your truly innovate and unique product. Step 3, get acquired by a large company and then shut down because your product competed with theirs.

End result, general public possibly benefited from your innovation, but in the long run, it was temporary.

Maybe the incentives would be better if it were harder for large companies to acquire small ones? If the path to riches where driven primarily by delivering value to customers. Would love to hear other's opinions on this.

> What are the outcomes though?

NVIDIA, and contractors who build data centers, and manufacturers who supply them, will all get rich.

The new kids have an easier time focusing. the big kids can integrate AI with their existing products and user data

In the long term, big kids win no? The big kids are also going to have an easier time with hardware at scale too

"but for some reason life is better in EU" citation needed

>It looks like US is winning but for some reason life is better in EU and innovation is faster in China

As measured by prosperity life in the US is better; the poorest US state has a higher GDP per capita than most western European countries. Americans have bigger houses, more food, bigger cars, bigger salaries, and access to better medical care and schools if they've got an okay job. Most Europeans are lucky to make $40k/year post-tax. And America is still winning on innovation because its AI models are ahead of China's both in benchmarks and in user preferences. How many people do you know professionally who use a Chinese model and agent framework instead of e.g. Claude Code or OpenAI Codex?

Claude is clearly the most superior product right now.

Gemini is absurdly expensive for low quality (3000 USD of tokens are not even worth what you get @ Anthropic for 200 USD).

Yeah I’m never using a Google product. The sole purpose of their company is to be evil. At least other companies are indifferent.

How does their top tier subscription compare in usage limits to the $200/mo Claude usage limits?

Another basic feature that’s missing is sharing a Gemini chat as a link anyone can view.

OpenAI figured this out: it’s awesome marketing when people send each other links to the app with a convenient text box to continue the conversation. It’s viral.

Google meanwhile set this up so that “anyone with the link can view” is actually “anyone with the link and a Google account”.

That’s grade A failure of marketing.

The PM in charge of that decision ought to be walked off a plank.

Google has barely released a successful product in 20 years.

well, it's basically existential, so the incentive is there to not only get it very right but also to limit the delta with how right anyone else gets it. The same can't really be said of the long tail of products Google have done.

Look to GCP as an example. It had to be done, with similar competitive dynamics, it was done very well.

Look to Android as another.

Because Google has the money to build 10 different versions/iterations of Gemini and can essentially force one to work. They have most people's data and most people use them for mail/search/browser/maps as well.

In my opinion though this is a race to the bottom rather than a winner takes all situation so I don't think anyone is coming out ahead once the dust settles.

Agree. Look at how miserably MSFT has failed at integrating AI tastefully in their business.

Google makes money selling ads. Nothing else matters.

Because the product quality doesn't matter if the competition isn't making any money.

google the only ai which invests mixing llm ai with real ai, and it seems work well.

race to the bottom. google in house cheaper inference hardware. anthropic buys it.

Persistence. Google has a lot more endurance then OpenAI does in this game.

The current AI market is going to destroy anyone who's specialized into it compared to having alternative revenue streams to subsidize it.

The conclusion Google is engaged in consumer capitalism is wild.

They're engaged in computing research and merely engage in consumer capitalism as a consequence of political and social constraints.

Products are a means to an end not the goal.

OpenAI and Anthropic are product companies and are more likely to fail like most product companies do as they will lack broad and wide depth.

Google has experience in design, implementation, and 24/7 ops with every type of SaaS there is. They can bin LLMs tomorrow and still make bank. Same cannot be said for OAI or Anthropic.

Do they though?

Google does things I hate with their products. But the money printing machine keeps going whrrr faster and faster.

Revenue, but what about profit? Google can be cash positive but I’m not sure Anthropic can be the same.

Google is playing the datacenter game differently because they have their own hardware.

> they know the risks of being dependent on Google

I wouldn't argue it's that risky. Look at their past entanglements:

1. Google Default Search Bribe - brings in $20B a year for literally doing nothing

2. Google Maps: Google let them build their own custom app using Google's backend, and it worked fine all the way up until Apple chose to exit that arrangement

actually I can't think of any others, but is there an example of Apple getting burned by Google?

[delayed]

this matches my experience. i'm building a mobile game solo and the amount of leverage i get from claude is wild, probably saves me 15-20 hours a week on stuff that would've required either a second person or just grinding through slowly.

$200/mo is nothing compared to what that time is worth. and it keeps getting better with each model release, which is the opposite of what usually happens when you pay for developer tools (they get acquired, enshittified, or abandoned).

the meta point about this funding round imo: competition between anthropic, openai, and google is the best thing happening for small builders right now. it keeps the tools improving fast and pricing competitive. if any one of them had a monopoly we'd be paying 10x for worse output.

I think GP is probably implying that this particular vertical requires obscene amounts of capital to keep up, which makes it really hard for a startup if you’re going up against businesses with giant free cash flow machines.

It’s the same reason Reid Hoffman sold his AI startup early… he realized he just couldn’t beat Google/FB/MSFT long term if it devolved into a money race.

Look at Sundar's most recent remarks and tell me Google's isn't only focusing on AI: https://blog.google/company-news/inside-google/message-ceo/a...

Basically "we have youtube subscribers" is the only thing that isn't all about AI, but even that i'm sure they're trying to figure out how to shoehorn AI into that product

Also Theranos was do-or-die and we know how it ended.

Better be otherwise it’s a 27+ year wait for breakeven!

Pay attention to the outflow of tech investment in the stock market. That money is going to move into OpenAI and Anthropic IPOs. The valuations will be as big you are thinking because the market believes these companies will represent an entire basket of startups.

Don't worry about it: they'll make it up in volume

https://www.youtube.com/watch?v=CXDxNCzUspM

yeah it's a slippery slope forcing companies to go public at X valuation. who decides that? what number makes sense? etc. but i do think we need to somehow fix massively overpriced companies going public and dumping on retail

I doubt talking heads on the TV can move markets.

His newsletter (and podcast) are fantastic.

>What was their original mission?

Beat OpenAI. The Founders came from OpenAI so there was obviously some disagreement about the direction there or they simply wanted more control.

To maximise shareholder value.

Friendly AI

Eh, I think you made the best decision you could given the info you had.

I’ve poked around on EquityZen and was shocked at how little information is available to investors. In some cases I did not even see pitch decks, let alone one of the first companies I looked at had its top Google result: CEO recently arrested for fraud and business is almost worthless now.

Unless you are willing to take a blind punt or have insider information, those platforms are opaque minefields and I don’t fault you for not investing.

Matt Levine has a fun investment test: when presented with an opportunity, you should always ask, “and why are you offering it to me?”

Meaning, by the time it gets offered to retail investors (even accredited ones are retail) we’re getting the scraps that no one else wants.

what are the private investor platform you mentioned ? and what are the requirements to join in?

A few years back, well ok maybe almost ten now, but regardless- a recruiter reached out to me about a role at a "series G" company like it was a selling point, and I was just kinda like ok maybe thats signaling its relatively stable and can raise money, but at the same time, that's a lot of rounds to have preferences ensure unprivileged shareholders get nothing, and also to have most of the hockey stick growth already tapped out.

This was in the middle of the boom when companies were fighting over talent, so I found it odd.

It's wild watching people fall for it

What’s your max extent? I was just doing some napkin math to think about where they’ll cap out.

China might purchase the data and train their models just to make the AI bubble pop. A few billions to throw a wrench in your competing superpower economy might be totally worth it

And yet open models have been tailing closer lately?

60%+ margin on inference: source ?

+ r&d costs

Of course, if one does not "pay" for investment, benefits are easily made ..

I am struggling with this because I have an Anthropic offer vs another equivalent offer that is all cash.

But project out forwards.

- What happens when Google builds a similar model? Or even Meta, as far behind as they are? They have more than Anthropic in cash flow to pour into these models.

- What happens when OSS is "enough" for most cases? Why would anyone pay 60% margins on inference?

What is Anthropic's moat? The UX is nice, but it can be copied. And other companies will have similarly intelligent models eventually. Margins will then be a race to the bottom, and the real winners will be GPU infra.

Is their overall margin also about 60% too? Or something saner like 30%?

The collective noun for a group of unicorns in AI is known as a hallucination, as in: a hallucination of unicorns.

I haven't used it to replace workers though, only to replace Google search. My company is pushing copilot but it's only $16/user/mo. Hardly lucrative and no moat.

Google has an edge, always a "personal experience" comment about leaving OpenAI/Anthropic in the dust every time new model gets posted.

It is approximately nothing, since lots of MS engineers use Apple products too.

Anthropic's marketing somehow punches hard. Not sure why, but the stuff they do sticks. Not because the products are great, but because the way they communicate about it gives people the right feeling. They do have legitimately the best coding model now for most tasks, and for narrative prose, but the marketing stuck and people stan'd them even when they were trailing.

The moat seems rather small right now. There are 7 different companies represented in the top 10 models on openrouter.

Couldn’t their excellent model and coding experience generate another excellent coding CLI tool?

> tens of billions of dollars to train

Source??

> cutting edge model that requires tens of billions of dollars to train

seems like there are a lot of those out there these days, and the costs are falling

> a massive concentration of talent and experience

Apparently 3000 employees? There's plenty of talent to be found elsewhere. Plus employees can be hired away.

> brand

meh.

> one of, if not the best, coding experiences

Seems easy enough to replicate, given how quickly they built it.