Hacker News new | ask | show | jobs
by voidspark 422 days ago
> everyone who has so far proven unable to compete

The article explains that Nvidia's biggest customers (50% of datacenter revenue) are switching to their own hardware.

5 comments

I'm skeptical of this. They have been doing this for a decade already. So far, those same companies have just become bigger and bigger customers of NVIDIA.

NVIDIA is very strategic about building product to avoid commodification -- both by building out network effects where software is tied to their proprietary sdk libraries, and by always focusing on being at the cutting edge of product.

Both these things can be true: a large company should try to build their own hardware to reduce supplier risk, and a large company should be open to suppliers that have better product that delivers business value.

So far, these large companies' internal hardware has been useful internally but never a complete replacement for NVIDIA, which keeps staying at the cutting edge of new capabilities.

NVIDIA already faced existential risk when Intel was commodifying all the dedicated motherboard components in the late 90s, 2000s, (like sound cards etc), so they're hyper-aware of this.

> So far, these large companies' internal hardware has been useful internally but never a complete replacement for NVIDIA

Google use their own hardware for AI/HPC. Nvidia hardware is offered to external customers who demand it on Google Cloud Platform.

The AI datacenter explosion has only occurred within the past two years. We are talking about plans that will take years to implement. The hyperscalers are trying to cut out Nvidia as soon as possible.

I'm baffled seeing this. I'm absolutely convinced of Google value prop, knowing Gemini 2.5 pro was trained and inference is performed on their TPUs is enormous. It's far and way the best model (Claude still has my heart, but their usage limits and limited context can't compete).

The company that can allow collaboration up the value chain only like Google can in this space right now is going to win. The author went over similar pushes by Meta, AWS, and Microsoft.

Now that their eyes are on the prize, and the prize is so staggeringly big, I'm convinced of the threat to their moat.

Google's Gemma new QAT training comes from google using their own TPU's

don't know much about the rest tho

I agree. In the exact same way that hyperscalers now build some of their own network stuff, they still buy an absolute boat load from traditional (specialist) vendors.
But then they're not boosting NVidia's competitors either, so wouldn't NVidia stay in the top position in their market ?

The article seems focused more on stock price and where to bet, than the market for GPUs or generic hardware vendors.

> But then they're not boosting NVidia's competitors either, so wouldn't NVidia stay in the top position in their market ?

Nope; if hypothetically 100% left NVidia, whether to their own hardware or to not use GPUs at all, it'd be easy to say NVidia would be last in the market

Are you arguing that the major cloud providers getting away from consumer GPUs would put AMD or Intel ahead of Nvidia in the GPU market ? How does that happen when the next biggest market (gaming) is still fully Nvidia's turf?
> Are you arguing that the major cloud providers getting away from consumer GPUs would put AMD or Intel ahead of Nvidia in the GPU market ?

Just the hypothetical "if X% of your customers leave, but don't go to competitors, won't you keep your relative market position"

He's talking about datacenter GPUs which are not really GPUs.

Gaming is not entirely Nvidia's turf either. AMD has 75% of the console market (Xbox and Playstation) and 30% of the PC market.

> AMD has 75% of the console market (Xbox and Playstation) and 30% of the PC market.

Looking at these numbers[0] for the console market:

> Sony has sold 61.94M units of PS5 and Microsoft has sold 30.14M units of Xbox Series X|S and Nintendo Switch sold 143.49M units.

Nintendo (NVidia chip) sold 1.5x of Sony and Microsoft combined. Given Switch's success the numbers look reliable to me, then Switch 2 is of course also from Nvidia, and I wouldn't be against it selling well.

[0] https://hookedontech.com/switch-vs-ps5-vs-series-x-console-g...

Old numbers. PS5 has sold 75 million

https://www.ign.com/articles/ps5-has-best-holiday-ever-overa...

In terms of revenue, AMD has the vast majority of the console market.

Yes the Switch sold more units, but that is a smaller SOC with much lower revenue.

Revenue and profit is what matters, not the number of units sold.

Check the price of the PS5 Pro vs Switch.

AMD has dominated the console market since Xbox One / PS4.

Nvidia could be earning $30 - $50 per Switch, while AMD earns $120 - $200 per PS5/PS5 Pro.

By switching to their own hardware, they become NVidia's competitors.
That's kind of arguing that people cooking at home are restaurants' competitors.

There's a kernel of truth in it, but if I was McDonalds I'd care a lot more about what KFC is doing than the market trends of cast iron pans.

I'd say so, yes. If I'm McDonald's and me and kfc are down, but Ralph's is up, I'd suspect a worrying trend about my industry.

It's kind of like how people may make fun of YouTube seeing TikTok as a competitor. But you dig deeper and realize why they decided to get into short for content.

Bad analogy

90% of Nvidia's revenue is from datacenters.

If the datacenters stop buying Nvidia's products, and use their own hardware instead, then Nvidia loses 90% of its revenue.

> Nvidia loses 90% of its revenue.

These 90% will be a flash in the pan, same way the COVID revenue was for mask sellers. Sure it feels bad from Nvidia's perspective, but we also can understand that the AI boom would not have kept Nvidia skyrocketing infinitely anyway.

No, you wouldn't, not if that market trend causes your revenue to plummet. Changes in the size of the market matter just as much as market share.
The 3, multi-trillion dollar hyperscalers using their own GPUs are individuals cooking at home? And Nvidia is the “restaurants”?
I'd say yes. We see the same dynamic with megacorp building in-house kitchens and paying staff to feed their employees, who stop going into restaurants at lunch.

It's a huge shift, but not something that can be acted by Nvidia, nor something they had to care about 5 years ago, nor a primary concern 5 years from now. On the long scale, it's almost as if the big fireworks that pushed Nvidia to its curent valuation just disappears and they're back to selling GPUs to OEMs and consumers primarily.

And something like 90% of Nvidia's revenue at this point is from the datacenter market.
As a gamer, I selfishly wish that Nvidia would go back to being a gaming company instead of a crypto and AI compute company. I miss MSRP graphics cards :(
I also miss sane MSRPs... I thought computers were supposed to get cheaper over time, not more expensive.
That stop happening a while ago. Newer chip are more expensive, as with memory.
Gentle reminder a lot of mid - lower end graphics card are far from MSRP. And arguably it is crypto and AI compute on server that partly subsided some of the R&D cost on consumer GPU.
You linked to an article about Google renting an insignificant amount of additional capacity.

Google runs AI / HPC workloads on their own hardware and has been doing that for more than a decade. Google Gemini was trained on TPUs developed in house. It does not run on Nvidia hardware.

And I believe Apple refuses to use NVidia as well, they’re actually using Google’s TPUs.

https://www.tomshardware.com/tech-industry/artificial-intell...

Apple is definitely using Nvidia hardware.
That was for training. For inference, they reportedly use their own silicon.
There was a rumor that emerged a few weeks ago that they broke down and made an order from nvidia:

https://finance.yahoo.com/news/apple-might-ai-game-1-1951003...

Before that, in a Wired article from 10 years ago about siri and ai, one of the apple higher ups was quoted bragging about having one of the baddest gpu farms around (paraphrasing)

Google, last month: “we’re doubling down on our partnership with NVIDIA”

https://blog.google/technology/ai/google-nvidia-gtc-ai/

That's Google Cloud Platform. Of course they will provide Nvidia hardware as demanded by external customers.

But their internal workloads and their frontier model (Gemini) runs on TPUs.

Yes, but the cloud customers "who finance TPUs" have NO INTEREST in TPUs and in Nvidia GPUs instead.

How does Google pay for TPUs internally? By Google Search and Google Cloud of course. Google Search uses TPUs, Google Cloud however has way more non-TPUs instances.

What people forget, nobody wants to switch from CUDA dependency to SW dependency on Google/AWS/Azure. CUDA at least allows me to use it in consumer, in pro HW, in cloud and AND in on-prem data center.

I'm really looking forward to Fortune 500 companies sending all their internal company data to Google to structure it to train custom AI models. Yeah, that will never happen. What happens instead is that Fortune 500 companies will build up AI expertise to build their own custom AI model and they will think hard if they want the training AI compute internally or on a cloud. Nvidia has a huge business of building data centers on-premises which people totally oversee. NO CSP will ever compete there because it's against their primary business model. A Reliance India contract from 2023 alone is a delivery of 2 million GPUs in a few years. That's probably more than Nvidia's last year's total revenue and that is 1 large corp in India only.

That’s the fundamental premise of the article: The hyperscalers will consolidate GPU compute exactly as they consolidated all other forms of compute. Including highly sensitive compute like product design and customer databases.

You can argue they won’t, but the “enterprises won’t put sensitive data in Cloud” ship sailed years ago.

> I'm really looking forward to Fortune 500 companies sending all their internal company data to Google

Their internal company data is already on cloud servers. They’re not going to waste money on doing it all in house. The executives will buy the AI service from Google/Azure/AWS, where the company data is already hosted, avoid the costs and risk of doing it in house, and collect their bonus.

NVIDIA ain't spent much time in the NFL else they would've known "...when you’re bleeding a guy you don’t squeeze him dry right away. Contrarily, you let him do his bidding suavely. So you can bleed him next week and the week after at minimum."