Just like the Nvidia RTX 5000 series launch, so-called “paper launches” are going to become more common

Nvidia’s RTX 50 series has been dubbed a “paper launch” by many, in that it’s been seemingly impossible for most normal people to actually get their hands on one of the cards. Unfortunately this trend has been going on for quite a while across the tech industry, upwards of twenty years, and it’s likely only going to get worse in the future.

We’ve seen “paper launches” for more than twenty years

ATI and Nvidia were accused of it in 2004

The reason a so-called “paper launch” happens is largely due to supply constraints. Production of high-end GPUs is a complicated process, and Nvidia’s 5000 series is made on a customized version of TSMC’s 5nm node, which the company calls 4NP (not to be confused with N4P). When silicon is produced, the success rate in production is referred to as the “yield”, but even those that fail can sometimes be stripped down and repurposed into something else. While TSMC’s 5nm process should have a fairly high yield, it would appear that there are still stock issues when it comes to the RTX 50 series.

According to Moore’s Law is Dead, even Nvidia employees are struggling to pick up the GPUs, which could normally be purchased through internal channels at the company.

“Not sure how widespread the dissatisfaction is, but I • can at least tell you anecdotally that I [Nvidia Engineer], and every coworker I’ve spoken with on the matter is outraged with how this launch has been handled. The employee storefront states that “due to shortages” no RTX 50 Series supply will be available until “widespread stabilization”. This hasn’t happened before, everyone is confused, and there’s a lot of people asking questions even inside Nvidia as to why we’re screwing over gamers so bad this time around…”

Where those stock issues are coming from isn’t clear, but we do know that Nvidia has become more of an AI company first and a gaming company second. As well, more advanced technology is harder to produce, and it could be the case that the company’s custom TSMC node might be struggling to meet the demand required to get these GPUs into the hands of consumers.

This isn’t the first time Nvidia has been accused of a paper launch, but it’s the first year where this is genuinely a problem that even affects staffers at the company. This has been a problem for so long that ArsTechnica published an article in 2004 titled “The Paper Launch: NVIDIA and ATI’s release circus“. A quote from the article:

More recently we’ve seen this in the video card business. ATI and NVIDIA have been trading punches for quite a while now, but their last round of product updates have left a sour taste in the mouths of many. The Radeon X800 XT Platinum Edition debuted in the beginning of May, but good luck finding one. Why release it early when, nearly two months later, most companies are still reporting a two-week waiting time (some say they will ship Aug 1, but many of these also said they would ship in June, too)?

You may recall that NVIDIA launched its new NVIDIA GeForce 6800 Ultra in early April. The X800 XT PE was simply a marketing trump on NVIDIA, who itself released its new flagship card too early, likely to steal thunder from ATI. The 6800 Ultra is also practically impossible to come by in the retail chain, unless you buy it with a complete system from the likes of Alienware. There’s a few places that claim to have cards, but they also want twice the MSRP for them, too.

Sound familiar? Funny how, just over twenty years later, we’re still making the same criticisms of the GPU industry today.

Related

Here’s how TSMC protects itself from earthquakes

Following a tragic earthquake in Taiwan, we looked into how TSMC protects itself from earthquakes in a place that’s prone to them.

These problems will only get worse

And they’ve been getting worse for a while

As technology becomes more advanced and companies race to get a leg up over competitors, we’re likely going to see these kinds of launches become more common. With the vast majority of Nvidia’s revenue coming from AI, it’s not even that big of a deal if it fails to sell enough units to regular consumers, as it’ll continue to make money from AI workloads that companies have been pumping money into.

On top of those newfound priorities for companies like these, there’s also a question of demand and production capacity. With more people than ever interested in high-end GPUs, lower yields on higher-end silicon, and more demand from big companies that want the same technology, there’s a lot of competition to fight in order to get the silicon that you want in your PC. The RTX 5090 that you’ve been vying for has silicon inside of it that could have been used in the process to make one of the company’s coveted H100 GPUs, and those sell for around $25,000.

Because of pressure on supply, it’s likely that the problem will only get worse; and affect more than just GPUs. We’ve already seen chip shortages in the wake of COVID-19, droughts in Taiwan, cryptocurrency, and a US-China trade war (amongst other things) caused a massive shortage of tech across the globe, raising prices and making it hard for consumers to get anything from smartphones to GPUs.

Unfortunately, we run the risk of similar occurring again, and this time, there won’t be a clear path forward to get ahead of it. Companies will likely continue to struggle to meet demand as time goes on without serious commitments to improving production capacity, We’ve already seen the AMD Ryzen 7 9800X3D fail to meet consumer demand, and it’s unlikely to be an isolated problem.