alt Hacker NewsAmazon has mostly sat out the AI talent war
Comments
Does Amazon want to be an AI innovator or an AI enabler?
AWS enables thousands of other companies to run their business. Amazon has designed their own Graviton ARM CPUS and their own Trainium AI chips. You can access these through AWS for your business.
I think Amazon sees AI being used in AWS as a bigger money generator than designing new AI algorithms.
That's because there is no lock-in in the current ecosystems for AI. Yet. But once AIs become your lifetime companion that know everything there is to know about you and the lock-in is maximized (imagine leaving your AI provider will be something like a divorce with you losing half your memory) these parties will flock to it.
The blessing right now is the limit to contextual memory. Once those limits fall away and all of your previous conversations are made part of the context I suspect the game will change considerably, as will the players.
> Of course, the AI talent war may end up being an expensive and misguided strategy, stoked by hype and investor over-exuberance.
To me, that's a pretty good explanation.
The world is crazy with AI right now, but when we see how DeepSeek became a major player at a fraction of the cost, and, according to Google researchers, without making theoretical breakthroughs. It looks foolish to be in this race, especially now that we are seeing diminishing returns. Waiting until things settle, learning from others attempts and designing your system not for top performance but for efficiency and profit seems like a sane strategy.
And it is not like Amazon is out of the AI game, they have what really matters: GPUs. This is a gold rush, and as the saying goes, they are more interested in selling pickaxes that finding gold.
AWS specifically have really dropped the ball on this.
I interact regularly with AWS to support our needs in MLOps and to some extent GenAI. 3 of the experts we talked to have all left for competitors in the last year.
re:Invent London this year presented nothing new of note on the GenAI front. The year before was full of promise on Bedrock.
Outside of AWS, I still can’t fathom how they haven’t integrated an AI assistant into Alexa yet either
AWS has always ridden other products (Postgres, MS-SQL, Redis, etc) that are open source or has negotiated licenses (Windows, MS-SQL, Oracle RDBMS) that are bundled in the end-user price per hour/GB/whatever.
AWS has Bedrock to use various AI providers and has bundled the licensing into the price, so they are getting the users without having to develop the actual AI.
They provide the compute, networking etc, and they provide the users to the AI vendors.
Why would they need to develop their own?
I do not see it as a master strategy. It is just something that happened. Similar to having no plan of having million fakeish Chinese brand selling crap on Amazon.com. But ones they are there then Sure, why not . May be in few year a lot of crashed and burned AI talent will be looking for boring corporate IT/AI job and Amazon will be around to offer that. And if does not happen there will still be ton of other work to do for Amazon.
Amazon seems to be taking the "When Everybody Is Digging for Gold, It’s Good To Be in the Pick and Shovel Business" approach here.
Don't need to train the models to make money hosting them.
I like how the author threw this in as (nearly) the last sentence:
Of course, the AI talent war may end up being an expensive and misguided strategy, stoked by hype and investor over-exuberance.
Isnt amazon basically Anthropic's HW partner very much like OpenAI has microsoft ?
I don't know who needs to hear this but, you can be a big tech company and not compete for every single market the other big tech companies are going for.
creative talents needs more comfortable and unconstrained working environment.But that against and broke many Amazon‘s leadership principal. Also amazon dont want to pay more money to those expensive ai talents.
The back-loaded vesting schedule is such blatantly cynical bullshit. It shows that they're planning to overwork you, push you to wash out, and undercompensate you for the experience, which is exactly what I've seen happen to a good number of friends. Amazon has become notorious here in Seattle - everyone knows they're a burnout factory. Some people make it through, and they make good money, but you have to really care about money for that to be worth the effort.
I had an Amazon interview loop on the calendar during my recent job search, a couple of months back, but it was difficult to get excited; they think so very highly of themselves, for what they're offering - and I don't just mean the money, but the culture too. They treat you like an interchangeable wage slave, not like a respected professional; it's all hoops to jump through, and procedures to memorize - dance, monkey, dance!
The recruiter was shocked when I cancelled the rest of the interviews, like, aren't you even going to give us a chance? But no: I had received a good offer from an ambitious, well-organized, well-funded AI startup which was excited to have me on board. With that on the table, why would I put up with Amazon? They won't offer better pay, they can't offer a better culture, and they don't have more interesting problems to work on.
Sounds like a winning strategy and a money saver to boot.
If you are a top AI researcher, there is no good reason to go to Amazon. For what? Pay? Career development? Company prospect? Work-life balance? You get nothing compared to what other companies offer.
And I say, good. We need new, smaller companies with different cultures in this space. We don't want these giant corporations to dominate and control everything.
Amazon would have a lot of trouble competing for top talent due to their reputation.
Amazon also didn't read the room when it fired most of its Alexa staff just as GenAI was taking off.
https://www.cnbc.com/2023/11/17/amazon-cuts-several-hundred-...
Of course not being able to monetise Alexa has always been a problem, but these and the article's issues are all to do with poor planning and top tier business direction.
Of course they've stood it out. The rate of change and the R&D expenditure is off the charts. It buys them marginal utility to hire AI talent at incredible salaries to keep them at table stakes.
Meanwhile, the models are getting larger and more complex, with more users, putting the support infrastructure well beyond what individuals and even small companies can afford to outright buy. You can easily spend well over a million on even basic infrastructure to try to support some of the newer models and make it available to a few end users.
As a point of strategy for individuals and small entities, it really is cheaper in this case to spin up some AWS instances for a bit to do some LLM work and then spin them down when not in use.
So if you were AWS do you mine for gold? Or do you sell shovels?
from outside looking in i think this is a move i would support if i were in amazon leadership. let the other players pay for the AI movement, pick up the fruits of their labor a couple of years down the line. i dont think amazon's main play is AI anyways, if anything it's to facilitate AI with their complementary platforms in AWS
I am reminded of the Uncomfortable Amazon Truths ( https://news.ycombinator.com/item?id=20980025 ) by Corey Quinn.
While they're protected now, https://news.ycombinator.com/item?id=20980557 quotes the one I recall...
- Nobody has figured out how to make money from AI/ML other than by selling you a pile of compute and storage for your AI/ML misadventures.
Why would they? I hear their main revenue is in AWS and AdTech. Assuming this is true, why would they need bleeding edge AI?
> Amazon doesn't provide useful tools for building durable multi-AZ applications. Most customers are not going to implement Paxos
Don't really agree here, yes they screw you financially on cross-AZ bandwidth, but all of their popular services are built to work well across availability zones.
Most people don't need access to a low level consensus service like Paxos, instead they will be using one of amazons managed database services, or s3, that provides higher level abstractions, and automatically manages consensus behind the scenes.
It's probably trying to figure out what it can sell at a large margin.
AWS dropped the ball but they didn’t really try. Apple OTOH…
Well Amazon is mostly guided by pragmatism rather then than hype, so there they are, waiting for the dust to settle to see what directly helps their bottom line.
Amazon will reap the results of the talent war. Why?
Because Amazon will build services on top of the technologies that come out.
Just today on hn there was a guy that trained his tiny model and got better results than most of the big models. He wasn’t paid 200m.
The gold rush is here, but the results are still shaking out.
You have all these labs spending billions on researchers and training clusters, not seeing much return on investment, meanwhile Amazon just partners with the labs, and provides inference for their models, and that seems to be fairly profitable for them.
Why buy the cow if you can get the milk for free?
Amazon doesn't need people with particular credentials on the org chart to bring in VC money. They already have plenty.
That’s because LLMs are just snake oil… Look at what a flop ChatGPT 5 was. If someone manages to actually make something useful, Amazon has a stake in Anthropic. Otherwise why should they waste their money while competitors bankrupt themselves at scale over a hype cycle.
would have thought that Amazon's API-first internal systems would make it a wonderful case for AI. Unfortunately doesn't seem to be the case.
AI will subvert and destroy Amazon's internal management culture, where status is gatekept by who can write the best 6-page reports to read before the meeting!
Or more likely -- Amazon management knows just how hard writing actually is, how hard to produce something with clarity and signal instead of just common-knowledge cliches, and so they understand that this LLM wave is overhyped. They're letting the other big players do the hard work, and effectively selling LLMs short by abstaining from the race.
Yeah Amazon is massively struggling to hire due to the extremely bad reputation of Andy Jasshole and the RTO 5 policy, and this is not exclusive to AI talents, but is the case for every single role. We have had reqs open for a year in my team and nobody wants to join.
Truthfully, I don't think anyone would recommend their acquaintances to join Amazon right now.
That said, Amazon is actually winning the AI war. They're selling shovels (Bedrock) in the gold rush.
AWS is more focused on making money off the infrastructure than on the application itself. It took same approach with kubernetes and might I say it has been very successful.
They invested $8B in Anthropic so they will be OK.
Didn't Amazon aquihire Adept Labs?
Amazon, and AWS specially, just don’t have recognized leaders in this space at the helm. I think that’s OK as they should focus on the more boring but important infrastructure stuff.
Jassy’s long rambling answer on the last earnings call though does suggest that being way behind on AI is a sore spot for leadership.
Attracting top talent though is a challenge for Amazon beyond just AI. Their reputation has become a real issue and the top folks simply have better options.
> The company has flagged its unique pay structure, lagging AI reputation, and rigid return-to-office rules as major hurdles.
No mention of reputation for harsh/ruthless/backstabby management practices towards employees (including for tech white collar, not just biz and blue collar)?
Is that not a major factor? Or are they not aware of it? Or is mentioning it politically off-limits? Or is putting it in writing a big PR risk? Or is putting it in writing a big legal risk?
I know Amazon's reputation for treating employees poorly came up in multiple discussions at one university's big-name AI lab, for example. Not only do some people read the news, but people talk, in groups and privately.
Amazon just has to host LLaMa and Qwen locally, just as they do so many other packages developed by others, and charge for their AWS compute credits. Why do they need “AI talent”?
It is interesting but I think they're doing the right thing. AWS bedrock works pretty well and you can access frontier models plus everything open source on it. In the end, I was disbelieving that Graviton would be good but the latest r8g series are great for compute so I imagine GPU compute will similarly be mastered by them in time.
I had totally forgotten that I signed up for the Kiro waitlist. It seems Amazon has also totally sat out the interest in their AI offering.
Has anyone had a chance to use Kiro at all? At this point I'm not even interested in it anymore, even if I got an invite.
Eh, looks at my aws bedrock bill, I think they are doing alright.
No one is looking at this issue correctly. Saying out of the AI "talent war" is a smart move. AI is due to collapse under its own weight.
1) High-quality training data is effectively exhausted. The next 10× scale model would need 10× more tokens than exist.
2) The Chinchilla rule. Hardware gets 2× cheaper every 18 mo, but model budgets rise 4× in that span. Every flagship LLM therefore costs 2× more than the last, while knock-off models appear years later for pennies. Benchmark gains shrink and regulation piles on. Net result: each new dollar on the next big LLM now buys far less payoff. The "wait-and-copy" option is getting cheaper every day.
Um, Amazon has invested $8B in Anthropic.
I think they learned some hard lessons from Alexa.
The top researchers published enough details on how to build what works well. Amazon can copy what's useful. They'll probably do it in a way that makes profit, too. Neither talent wars nor AI, startup models contribute to that.
So what's the tldr, are they just too cheap too pay for top AI scientist talent-- which is imagine they would need in order to enter the fray?
> "GenAI hiring faces challenges like location
No! Really? With RTO? Unbelievable /s
[dead]
The evidence shows that there is no methodological moat for LLMS. The moat of the frontier folks is just compute. xAI went in months from nothing to competing with the top dogs. DeepSeek too. So why bother with splurging billions in talent when you can buy GPUs and energy instead and serve the compute needs of everyone?
Also Amazon is in another capital intensive business. Retail. Spending billions on dubious AWS moonshots vs just buying more widgets and placing them across the houses of US customers for even faster deliveries does not make sense.