Twitch's profitability is an accounting decision. Twitch's IP was transferred to AWS and became IVS [1]. IVS is sold to other companies like Kick. It's a completely subjective decision of how much Amazon charges itself for use of IVS by Twitch. If Amazon lowered the IVS price it charges itself, Twitch would suddenly be hugely profitable.
All this shows is Twitch is not an Amazon priority.
>All this shows is Twitch is not an Amazon priority.
I'm pretty certain Amazon didn't buy Twitch so they can subsidize them and effectively run them at a loss. The point is for Twitch to make Amazon money, not the other way around.
I said that whatever price AWS charges Twitch, it makes Twitch unprofitable. Since this is all just moving numbers around on a spreadsheet, they could choose a different number and Twitch would appear to be profitable.
The point is that Twitch being unprofitable is an artifical narrative not an objective reality.
I'm aware. This just further makes my point. Twitch develops IVS. The internal accounting is that numbers move around in a spreadsheet to "pay" for that. Amazon then moves more numbers around to "charge" Twitch for IVS, AWS or both.
You can select one set of numbers that makes Twitch look highly profitable or highly unprofitable with zero bottom-line (to Amazon) difference.
Why I bring this up is that Twitch uses it's unprofitable status to both justify layoffs and to reduce what they pay to creators. They've cut the sub split. Who knows what the ad revenue split really is. AWS is a huge profit center for Amazon. The fact that IVS continues to exists strongly suggests its profitable. If not it would be cancelled. But that profit is being attributed to AWS not Twitch even though its Twitch IP and Twitch employees who work on it.
My point is a completely fabricate narrative is being used to cut costs and thus increase profits.
As I saw it, still think this is in reverse. Twitch as an org was running at a loss and had some tech and network that could be packed up and sold to enterprises.
AWS, which is the B2B side of the house, resells this as IVS, take some % to pay for sales and integration. Twitch makes a profit on IVS, but IVS alone doesn’t pay Twitch bills. Now the Twitch org runs less of a loss.
This is pure partnership play. With the advantage that the partner is in-house, integrated on the marketplace and can rely on the house sales team as well.
I don’t believe that the tech is not profitable. It’s just that the business of making ad-supported B2C profitable is hard.
If I were to guess, I’d say AWS IVS customers are a really tiny % of Twitch’s traffic, but are the most profitable.
I would expect there to be legal requirements around anticompetitive behaviour.
Even if not, your comment seems to imply (to me at least) that the vast majority of Twitch’s operating costs comes from paying AWS (implied in this is the markup for AWS, not the actual costs of running the infra).
I don't understand your point. It seems that you are saying that this is all an accounting trick by Amazon, but we don't really know whether they "charge" Twitch anywhere close to the costs that it actually incurs.
Over-hiring adjustments is something I think was more common last year. By now, all companies "needing" layoffs to compensate should have already done so.
This feels more like Twitch just really struggling financially, as mentioned in the article.
For Twitch in particular I think this is a convergency both over-hiring and post pandemic business reduction.
Twitch got really big during the pandemic: lots of people had nothing else to do besides sit on their computers all day playing or watching games. Viewership exploded, creatorship exploded, hiring was easy in remote, so hiring exploded.
Post pandemic Viewership has likely tanked as people got back to work and school. Creatorship has also subsided as people can't play games 24/7 anymore, so any revenue growth they would have experienced over the pandemic has been likely off-set by now.
They countered that with layoffs last year, as most tech companies did, but their case was not just about laying off the extra hires, they also lost business they could not recoup, so they're downsizing even more.
Over-hiring and restructuring explanations, i.e., excuses, are the most lame and slaps in the faces ever. Such reasons are solely management problems and due to management incompetence. I feel that regulation could be helpful here. The employer's side of at will employment should be reduced and other pressures to avoid these situations.
Overhiring always being a mismanagement issue due to incompetence is a cute narrative for the labor-vs-capital arguments, but it's not literally true unless you believe that failing to read the future accurately is mismanagement. Management's job is to do the best they can with the forecasts they have, and no reasonable person believes that forecasts can be accurate every time.
You can rephrase management's duties to "predicting the future" as a way to maximalize the so-called difficulties of their roles, but that's an unrealistic description. And it is in fact management's primary duty to understand the evolution of their organization and to have mitigation strategies. Simply hiring when you can and then firing at the first sign of trouble and when investors throw a hissy fit is about the most amateur, incompetent approach available. A monkey could perform that role better.
I was recently laid off just a year after being hired. The layoff was described as not cost cutting but restructuring. The reality is that no restructuring has happened, morale has tanked, and attrition in key positions has occurred. Me being laid off means that management couldn't even plan out a year. Further, this company had a restructuring (i.e., layoff) in 2020 as well. If you need to "restructure" every three years, you don't know how to manage.
At will employment is a poor setup because companies hold all the cards. They should have much more pressure on abiding by contracts.
Predicting the future is literally a core upper management job duty. It was fairly obvious and should’ve been assumed that interest rates would eventually rise.
Isn’t over-hiring a form of management incompetence?
Money was too cheap, tech companies are now slapped with larger debt servicing costs and a less captive audience. Growth focused and delusion forecasting done by companies weren’t grounded in reality or pragmatism. This led to things like over-hiring.
I have a feeling this is going to continue as AI / LLM's become more capable. I know I am late to the party but was messing around with chatGPT the other day to see if it could help me with a code issue I was stuck on. It blew my mind!
I don't see how it and similar LLM's don't lead to massive replacement of tech workers and creatives.
No? It will just widen the gap between code monkeys and actual developers who build scalable systems. It seems like these takes are only espoused by those with nieve or little to no experience in software development.
This is an odd thing to say when its actually happening in the real world.
People are getting laid off and replaced with AI.
I guess all of the leaders at those businesses though are" naive or little to no experience in software development."
I have actually been present in a meeting and watched in real time where someone demoed an LLM that absolutely replaced the need to keep several engineers.
There is no way to quantify "software development losses from AI" versus post-COVID hiring frenzy layoffs and cheap capital becoming more expensive. Any definitive statement claiming such is purely conjecture.
No companies who are doing layoffs are saying "we're laying off developers because AI", everyone who is drawing those conclusions are making assumptions. AI can automate other menial, clerical stuff, but making that stretch to software development isn't founded.
ChatGPT and friends coincidentally started gaining traction as the economy started cooling off. Rather than drawing casual inferences to AI, it's much more likely that companies are laying off because of the latter rather than AI.
> AI-related changes in how content is generated and shared
This has to do with them leveraging AI for their product, not replacing software developers writing code. They also didn't fire any full-timers, so I fail to see how this fits your assertions.
Not sure me saying "I have a feeling this is going to continue as AI / LLM's become more capable" is a grandiose statement; I think you may be overreacting but whatevs.
I'm sorry you are so angry friend, all the best. Wishing you happiness.
This would imply there is an “end of work”. If these things are so good at augmenting your work, why would Google not keep their engineers, augment them with LLMs and just be more profitable and productive with the same number of people ? What does the lay off part get them ? Why be “just as productive” if you could be moar productive?
The fully replacing people thing would be if we had an AGI in which case I’ll be using one to augment myself and stay on top of the game.
Because that's not how the real world works.
Any product has a real world limit on the amount of money it can generate.
Only so many companies are going to subscribe to product X.
There are only so many customers in a particular niche.
Your assumption is that generating more work = more profit.
The idea that just adding more engineers leads to more money is an obvious fallacy. Cutting engineers though and continuing to receive the same income is an obvious way to increase net profit.
Because that's not how the real world works. Any product has a real world limit on the amount of money it can generate.
Yes it is, I’ve never seen “the end of work”, have you ?
Google has an unlimited amount of issues to solve. I read about them here everyday. ChatGPT is even an existential threat to their search, if not directly, then through the internet becoming a huge piece of AI spam.
No one is talking about or implying the end of work except you.
What I am saying is that if a company can do more work with less people they will do it.
"Google has an unlimited amount of issues to solve" sure but that does not lead to profits otherwise they would just hire as many people as they could and make more money. Again that's not how it works.
We don’t if that’s the real reason or the advertised reason that they’re laying people off. If was an investor I’d prefer the "AI is making us money for nothing story". But I know blog spam is also screwing their product up. It’s why I pay for an alternative search engine.
Ironically AI seems like a huge problem for Google's main business. Even if what you're saying is true, and all the layoffs are because of AI productivity gains. Well guess what? Other people can use AI to build cheap advertising businesses too? AI is cheap and it's getting cheaper, thanks to open source AI, it's accessible to everyone.
What I've come to realize is that, it's not Youtube that's valuable, or Google, or even Open AI, it's the information these things are providing access too. The information people are creating for them. The communities.
Now it seems like accessing that information is becoming democratized in a way that Google never really planned for. Soon I might be able to have something that can generate a youtube video for any topic I need, without youtube, just locally on my laptop or phone. That's going to be a problem for them. They know this.
There is no evidence that any of your statements are correct. Yet somehow people keep parroting such comments. Is OpenAI desperate for customers to gaslight?
Without doubt, machine learning is here to stay. But it won't be anywhere near the promised value. People already realise they've been misled into thinking there's intelligence in it, and the narrative is changing. There's heavy backlash against ai support bots, ai generated content, and code is proven to not work as advertised. Furthermore, people are taking action against the rampant intellectual property theft and the less content to steal the less useful these tools are. Eventually the hype will stabilize and things will be back to a new normal.
Amortization is basically just a way of splitting the cost of something out over multiple years to change how much you may owe in taxes.
In this case, previously software engineers' salaries were considered an operating expense like any other salary, and the full amount of that was used to decrease how much profit you have to pay taxes on. Now you can only write off a portion of the salary, having to split it over multiple years' taxes instead.
This is really common with R&D expenses, physical equipment, etc. The big change was that software engineers are now considered an R&D expense rather than someone working to assemble a final product, for example.
As far as I'm aware amortization is only used for tax purposes. I guess you could use the adjusted profits and losses if you had a reason to show lower profits, but as far as financial planning goes the money has already been spent so its largely a game of when you want the expense to show up on your tax bill.
A lot of the advertising industry (Twitch is an ad publisher which uses tech, not a tech company) is moving away from high-volume low-intent advertising to low-volume high-intent. It's becoming harder to squeeze pennies out of banner and video ads on platforms like twitch, facebook, youtube, and google ads. Some retailers like Amazon and Walmart make a significant chunk (if not the majority) of their revenue from retail advertising (basically: promoted search results)
All ecommerce sites do this if they can; it's just difficult for the smaller sites to normalize product information and share user browsing data across DSPs. Grocery stores have been doing it for decades: when you see a display on the endcap at the end of an aisle, that's a product which has paid Target or Kroger for the prime real estate.
The company I work for is working on a retail media system so that smaller ecommerce sites can get paid to promote results by advertisers in the same way that they can get paid to show banner ads: the advertiser knows nothing about the retailer and vice-versa. The advertiser offers their products on our DSP and the ecommerce site makes their inventory available to us on our SSP and we connect the two
So in other words you are contributing to the enshittification of e-commerce, giving the smaller sites the same hostile customer experience as Amazon. I don't blame you, might as well make some money off the trend while you can. But I doubt that this will end well for the sites themselves. If they can't deliver a better experience than Amazon then customers will just stick with Amazon.
The way I see it (and the way I see all OpenRTB advertising) is that we are helping to give a competitive edge to smaller, more independent sites so that they can survive against huge corporations like Amazon. Without higher yield advertising, independent publishers and new organizations can't survive and ecommerce sites can't compete. Unless people start paying for subscriptions to certain services or choosing to utilize smaller businesses at a higher cost and slower delivery times (which they will do neither of willingly), the open advertising system is the financial engine which keeps the internet from consolidating into a walled garden of just a handful of sites.
> Twitch is an ad publisher which uses tech, not a tech company
That's highly disputable. In the first place do they offer a service for communities, making it a social platform. Then do they seem to work intensively with AWS and for AWS, making them indeed a tech-company too. I mean, AWS seems to resell Twitch-Technology since some years.
And finally did they had no significant ad-volume for many years, and in most countries did they not even serve any actual ads, outside of Amazon Prime Video-Ads.
Yes, they do work on transforming into an ad-company since 2, 3 years. But they are still on a pretty low level. Viewers are already complaining hard just because of 1,2 minutes of ads per hour. Getting to the level where they find their sweet-spot and making them a primary ad-company will still take some time, I think.
> is moving away from high-volume low-intent advertising to low-volume high-intent.
Yet, this seems to be where twitch is aiming for, because nothing else actually works with the type of content their customers serve.
I mean, they only added banners last year or so.
> the first place do they offer a service for communities, making it a social platform.
All social platforms are advertising publishers. Facebook is an advertising company (though they refuse to say so; calling themself a publisher is a legal liability for them), they make almost all of their revenue from selling ads. YouTube is an advertising company as well.
Working with technology (and even developing your own technology) doesn't make you a technology company. insurance companies employ software engineers, but they're still insurance companies.
Even if twitch fails to make any money at advertising, they're still an Advertising company.
> All social platforms are advertising publishers.
Yes, because originally, they had no other way to monetize, unlike Twitch. But Twitch has subs and donations for a long time now, and ads are just another stream of income for them. And the other platforms are now trying to add subs and donations to their income too, while Twitch is coming from the other side, and tries to expand their income through ads.
> Even if twitch fails to make any money at advertising, they're still an Advertising company.
Going by that logic, are they also a Tech-Company.
Tech companies sell tech. Advertising companies sell attention. Most big tech companies are just high tech advertising companies. They don't like to say this because it doesn't carry nearly as much prestige to admit that the culmination of their life's work is to get people to buy more crap rather than making the world better in any measurable way. I say this as someone who works for an advertising company
I take issue with pretty much every single one of your statements here.
> culmination of their life's work is to get people to buy more crap rather than making the world better in any measurable way
Advertising makes the world better in a lot of ways. It allows companies to provide services without actively charging for them which is a much preferred way of monetization of the vast majority of users (just look at the popularity of Netflix's recent ad supported plan). It allows companies and users to connect on new products which often results in a purchase that both the consumer and company benefit from. The result are higher living standards.
Google Search is funded primarily through ads. If you truly believe that Google Search hasn't made the world better in any measurable way I truly don't know what to tell you.
Then you'll probably disagree with me when I say that I think the world wide web in general has had a net zero impact on the world. I believe it's had as many positives as negatives. Perhaps if we could've frozen it circa 2005 it would be an overall positive but I'm not convinced it's as good as we all believe it is
They worry that if they consider themselves to be publishers of content, they will become liable for the substance of that content (e.g: calls to violence on their platforms). Thus, they go out of their way to specifically deny that they are a publisher.
> The oft-cited Section 230(c)(1) states, “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”3 In practice, this means online services are not liable for defamatory or otherwise unlawful content their users post. Section 230(c)(2) elaborates, stating that online services are not liable for “any action voluntarily taken in good faith to restrict access to or availability of [objectionable content.]”4 This second provision protects online services from liability for engaging in content moderation and enforcing their community standards.
It seems to me that amazon (and many others) do not comprehend that twitch streams are primarily advertisements. People voluntarily pay cash to watch voluntarily created advertisements in the form of someone playing a video game (the product that is being advertised).
Trying to inject shorter advertisements by pausing the longer advertisements makes no sense at all -- it reminds me of the "Malkovich Malkovich" scene from the movie Being John Malkovich. It makes unhappy customers (viewers and streamers) as well as distracts from the primary business.
Even if they didn't get internal discount, there's no way they pay retail as in advertised prices. At that size AWS itself would likely want bandwidth commitment for a longer period too.
