Why can't we back order a 3080?

lopoetve

Fully [H]
Joined
Oct 11, 2001
Messages
30,935
Just so you know, I'm enjoying our conversation. I don't mean to come off as insulting. Only being direct.

Brand Damage: Low inventory and high desire is one of the reasons 3DFX died.
Also issues with the product, the poor decision to get into the OEM space, conflict with their prior OEM partners, and a few other reasons (including poor timing of the market and various loans).
On a short term scalpers doesn't hurt your bottom line. On the very short term it bolsters it.
Nvidia makes no money from scalpers beyond what htey would make selling the product in the first place. Distributors (for better or worse, a scalper is another distributor layer) make the money there, not the OEM. Scalpers have no impact to the bottom line of the corporation directly.
But while short term profits are great for investor dividends and day traders, damage to the company reputation starves it of long term sustainability and erodes a customer base. Ask yourself, why the 3080 was priced at $699 instead of $999, and why did Jensen have to do a song a dance to 10 series owners? Why did they immediately have to enlist Digital Foundry to back up their claims? Because the Turing launch was THAT bad, and the pricing had eroded consumer confidence.
I'll set aside the company reputation portion for the moment - Turing was not great for hte home consumer (although we still bought a TON of them), but was absolutely massive in the datacenter space, AI/ML space, HPC space, and others. I've sold more turing cards than any other (excepting the M10 from the maxwell generation, as it has no replacement) for those use cases, by an order of magnitude - especially the T4 series of DC cards. Triply so when you include the fact that most of the world is WFH right now and there's massive drive for those cards for said initiatives. And those cards drive licensing revenue too, which is not only repeatable, but much higher margin than hardware.

As for the company reputation - so they've had a crappy launch. You going to only buy AMD cards now? It will be forgotten; human psyche these years is very short term focused, and then moves on to some new squirrel. We're also a somewhat captive audience as there are exactly two options right now - Nvidia and AMD, and part of the feature set is Nvidia only as far as we know and can purchase.
Had Turing been more successful, they would now be reaping far greater profits from what should be one of their most successful and financially strong generations. Instead they are selling for pitiful profits which inevitably provide less capital for both R&D (ask ATI about that) and further brand erosion (3DFX) by having consumers focus on negative instead of positive experience. Additionally, it has opened a massive door for AMD to step in if they play their cards right, and gut a hole into a previous locked up market (see AMD Ryzen).
They're making money hand over fist. They own a virtual monopoly in the DC/AI/ML space, as well as in the automotive space. There's also a balance to pricing beyond "maximize profits", as that definitely DOES damage your brand. And even with AMD playing their cards right - there's a lot of this space that they can't just "enter" as they simply don't have the tech yet (see again, DC space/etc, where AMD has effectively exited from the GPU perspective). Pricing is a highly complex subject, especially during a recession like we're in now.
The owners: Of course they're happy. Some of this is bolstered by the lack of availability though. They may even subconsciously be increasing their satisfaction by knowing they are part of an elite so to speak.

The folks who still want them: Most is the key word. Most means lost sales. Most means lost brand. Most means opportunity for competitors to steal customers. You know what locks people in? A non-refundable $50 deposit with the guarantee of delivery. Customers who make purchases on this level will often actively ignore the competition because they want to reaffirm their choice, and only associate within an eco-chamber of reinforcement.
Sure, but there are always some lost sales. What competitors? AMD (as far as we know) does not yet have a true competing product. Until Big Navi arrives, we don't know if it exists or not - or what it performs as. Lots of promise, lots of non-delivery in history - I hope Big Navi is good (want to upgrade my 5700XT), but we just don't actually KNOW yet. And even then, do we really think that it will be fully head-to-head competitive than the 30X0 series? Intel is not a competitor.

Those deposits open a HUGE pile of legal issues, as they are run differently in every state and across state lines - non-refundable isn't an option in some places, and rules around refunds are different between states/etc (see, for an example, the fine print on ticketmaster orders, and how it can differ between states).
Folks who don't want one: They ARE customers. Anyone that desires your product is a customer. The trick is to retain them until they have the capital to buy. Two things typically happen: They save up the money, or they purchase a lesser but similar experience. This is lost opportunity. Now AMD has a chance convert these customers into clients. How do you save that? Third party payment plan with a reservation and non-refundable but small deposit, followed by motivational marketing.

Tier-2 customers (folks actively avoiding you) are not actually customers. You concentrate on Tier-1 (ones who could be and want to be) and tier-3 (folks that don't know about your product somehow), while making sure Tier-2 doesn't do anything weird to you (consider Tier-2 the fanboys of the other side, in the normal marketing models) - https://cdn.slidemodel.com/wp-content/uploads/6539-02-blue-ocean-strategy-16x9-15-870x489.jpg gives a good example of the model normally used for this (this is partially my world). Again, this assumes that AMD's part can serve as a substitute - we don't know that yet, although you are correct, that many people will consider it a substitute. Then again, my RTX 2080TI is also a substitute - at least for now.
Angry at scalpers: A small but viable potential base that could convert to sales that may have otherwise missed. People will do all sorts of things when they feel slighted. Scalpers are notoriously hated. Nvidia could flip these potentials into sales by a reservation program and pushing a message of screw the scalpers. It's a small drop in a bucket most likely, but it's also extra motivation for all other potential clients.

Again, the model you suggest has traded short term profits for long term sustainability which in itself is a valid model IF you have a long term reward to reap from it and no or little competition. But they do have competitors, and they don't have much to reap from allowing their products into the hands of scalpers other than short term profits. If they continue to take that business model, then they risk eroding their public value and ending up on the bad end of a holding company sale interested only in their IP. This is part of the reason that Nvidia NEEDS Arm.
Sale was made - to the distributor (scalper) or end user, the effect to the company bottom line is effectively nil. You're right that in the long term, nvidia DOES care about scalpers - because eventually they damage the brand, or can, but for a single launch? With everything else going on? Eh. Impact is minor right now - they're selling as many cards as they can make, a community of people has risen to try and GET the cards, many of us are just shrugging and paying the scalpers (if we can - I'm staring at a Gigabyte 3090 on ebay right now for a reasonable price), and there's still not a competitor. If they find a way of dealing with this and getting parts into peoples hands over the next 6 months, everyone will be happy and it will be mostly forgotten before long.

Now, if Big Navi is truly competitive, and they can't figure out how to get yields up, or the scalpers find a way to totally dominate the market (profit margins there do matter to them, since they're effectively running on retail cost of goods sold), then they have to make a decision - Reebok/Nike/etc don't have issues with scalpers, as they've made the decision to accept that distribution model... Nvidia may not, as their products may not be unique enough for it to work or be valid (or the outcry might be significant enough to threaten the other lines of business), but until we see what their competition has, we just don't know.

Lots of unknowns here. It's far more complex than "just throw up a preorder deposit system", which is the point I'm trying to make - because that's not as easy as it sounds. Same argument I get into with folks doing rental of enterprise datacenter equipment - if your customer's accounting isn't really built around operational expenses, and is built around capital instead with either EBITDA or capital amortization as key, then it doesn't matter how cost effective your solution is - they're not changing their accounting model to buy a bit of kit. The CFO would shoot you.
 

Skott

Supreme [H]ardness
Joined
Aug 12, 2006
Messages
4,196
I dont see it getting easier to find the 3xxx cards until 2021 to be honest. The holidays are fast approaching so even more folks will be wanting to buy. I'll wait til next year when they roll out the Super or Ti model. Unless AMD has something good but I figure they will have the same lack of stock problems as the Nvidia cards. I wish these two companies would spend more time in production before announcing and releasing but it is what it is. Retailers now need to invest in anti bot programs as well. They shouldn't have to but its the world we live in now.
 
Joined
Nov 18, 2011
Messages
596
We have entered the Car Guys vs Bean Counters (by Bob Lutz) portion of this conversation; business philosophy. For those that don't know, Car Guys vs Bean Counters details Bob Lutz attempts to wrestle GM away from a short term fiscal profit driven philosophy that netted such gems such as the Pontiac Aztec and Sunfire, as well as many of the other objectively terrible vehicles of the 90s and early 2000s. Ultimately, GMs car models were so unsuccessful that it was their banking division holding up the company. When the housing market crashed, so did GM. It's an interesting read.

Portions of our conversation have actually gone into philosophical debate regarding business practice and models, flexibility, mobility, and especially ethics. We don't see eye to eye on those and can debate until the end of time but likely will never see eye to eye.

As such I'm going to wrap up my replies with this. Good conversation though.


Lopetve: Also issues with the product, the poor decision to get into the OEM space, conflict with their prior OEM partners, and a few other reasons (including poor timing of the market and various loans).

Which is why I said one of the reasons. It's not all of it but having one thing out of spec put more pressure on other components to carry the weight of the company.

Lopetve: Nvidia makes no money from scalpers beyond what htey would make selling the product in the first place. Distributors (for better or worse, a scalper is another distributor layer) make the money there, not the OEM. Scalpers have no impact to the bottom line of the corporation directly.

Scalping absolutely can and does of effect a companies bottom line by guaranteeing an early sales spike of possibly a poor product. It essentially bypasses the failed sell through of the first shipment entirely, and allow a company to present glowing reports to press and shareholders alike while giving time for correction. In the mean time, the reviews which were embargoed and launched after the sellout could point out a major flaw....such as a cooler that sounds like a vacuum cleaner (5000 series) or disappointments in performance.

Lopetve: I'll set aside the company reputation portion for the moment - Turing was not great for hte home consumer (although we still bought a TON of them), but was absolutely massive in the datacenter space, AI/ML space, HPC space, and others. I've sold more turing cards than any other (excepting the M10 from the maxwell generation, as it has no replacement) for those use cases, by an order of magnitude - especially the T4 series of DC cards. Triply so when you include the fact that most of the world is WFH right now and there's massive drive for those cards for said initiatives. And those cards drive licensing revenue too, which is not only repeatable, but much higher margin than hardware.


You are not wrong about the datacenter market, but you seem to be grossly underestimating the value of the consumer market in Nvidia's sustainabiltiy. In Q2 2020, datacenter revenue was $1.7 billion, which is up 167% from the previous year. However, gaming revenues were nearly the same at $1.65 billion but only up 26% from last year. This is because of the lack of Turing sales in the face of stiff competition from existing Pascal, and AMDs unexpectedly decent NAVI which scooped sales that would of otherwise gone to Turing. So despite the licensing revenue and other income sources associated with rise of datacenter sales, they only now under unique market circumstances have exceeded that of the gaming revenue which has a far less profitable after sale income source.

As for the other markets:

Virtualization was $203 million or down 30% year on year.

Automotive (which is not a monopoly to Nvidia), is down 47% at only $111 Million.

CPU and Networking revenue was up 130% year on year at $1.78 billion.

Overall, graphics was $2.08 billion of their company, or nearly 1/3 give or take a bit. They are trying to build their way out of it, but are still heavily dependent on it for a massive amount of their resources. With the accusation of Mellanox giving their datacenter revenue a massive burst, those networking and cpu growth spikes are unlikely to be sustained over the long term, and will likely drop down to industry standards.

To quote Jensen himself: "Mellanox grew sharply, driven by the need for high-speed networking in cloud datacenters to scale out AI services." Nvidia forced their way into this market and put their cards in their own product which boosted sales. It's a common and smart move. However, the growth rate is clearly not sustainable, and may attract regulator attention.

This is why I said Nvidia needs ARM. ARM is going to give them a forth pillar of profits which will absorb issues with consumer GPU side.

Lopetve: As for the company reputation - so they've had a crappy launch. You going to only buy AMD cards now? It will be forgotten; human psyche these years is very short term focused, and then moves on to some new squirrel. We're also a somewhat captive audience as there are exactly two options right now - Nvidia and AMD, and part of the feature set is Nvidia only as far as we know and can purchase.

Philosophical disagreement with you here. Humans psyche is very pliable but less short term then people realize. A single bad incident will be retold 10 times more than a good incident. Nvidia had multiple good launches but that doesn't draw attention because human brains are wired to look for problems or more precisely threats. As such, these events get seared into human psyche and it takes strong will power and knowledge to look past that.

Additionally we are in a very unique position in time. Right now the 3000 series faces not one but 3 immediate competitors with an additional on the horizon. Those are Big Navi, Playstation 5, Xbox Series X/S, and Xe (Horizon). Three of those threats are from AMD which is their biggest competitor and the one most likely to push into their GPU driven markets, except backed with a robust CPU division that has been firing on all cylinders. This is why Jensen attacked consoles in his presentation. Fortunately for him, the consoles got hit by bots as well. Unfortunately for him the consoles have a robust reservation system through nationwide retailers that Nvidia simply doesn't have, and is what started this whole topic.


Lopetve: They're making money hand over fist. They own a virtual monopoly in the DC/AI/ML space, as well as in the automotive space. There's also a balance to pricing beyond "maximize profits", as that definitely DOES damage your brand. And even with AMD playing their cards right - there's a lot of this space that they can't just "enter" as they simply don't have the tech yet (see again, DC space/etc, where AMD has effectively exited from the GPU perspective). Pricing is a highly complex subject, especially during a recession like we're in now.

Again the claim that they own a monopoly is a generalization that their profits just don't match. The Datacenter space grew because of their purchase of a datacenter company. Their AI division is growing but is tied into their datacenter sales. Their space and automotive divisions are declining and there are other competitors in this space including the major automotive manufactures themselves. Of course, should the acquisition of ARM become complete, some of the automotive competitors will be beholden to Nvidia as their CPU supplier.

Lopetve: Sure, but there are always some lost sales. What competitors? AMD (as far as we know) does not yet have a true competing product. Until Big Navi arrives, we don't know if it exists or not - or what it performs as. Lots of promise, lots of non-delivery in history - I hope Big Navi is good (want to upgrade my 5700XT), but we just don't actually KNOW yet. And even then, do we really think that it will be fully head-to-head competitive than the 30X0 series? Intel is not a competitor.

The simple numbers is that AMD has presented three avenues for NAVI products to enter your home and Nvidia Ampere has only 1. Those are Playstation, Xbox, and GPU. There is no reason to doubt early rasterization numbers from AMD due to the showing of the consoles and the absolute brand trust suicide it would be to post promises that it didn't achieve (aka Intel in the 9900K marketing fiasco). In addition, with the growth and good will that Ryzen has gotten, AMD is seeing growth that will drive more profits into R&D and right back into competing with Nvidia. Additionally, don't think that number 1 GPU maker in the world, Intel, can't come out swinging. Remember, every system that ships using Intel's or AMD's built in iGPUs exclusively is now a lost sale for Nvidia as they used to sell chipsets with the same functionality. Again, why they need ARM.

Lopetve: Those deposits open a HUGE pile of legal issues, as they are run differently in every state and across state lines - non-refundable isn't an option in some places, and rules around refunds are different between states/etc (see, for an example, the fine print on ticketmaster orders, and how it can differ between states).

Huge is subjective. It complicates things, but doesn't make them impossible, and there are always other methods to achieve the same goal. Nvidia spending $3 million (or 0.18 percent of quarterly revenue for consumer GPUs) in legal fees and workings to ensure customer accessibility is, IMHO, a small drop in the bucket for a company that gets approximately $1.65 billion in sales per quarter from that segment. The bottom line is that Nvidia has the ability and resources to establish this, even if its 5 times my estimate. Or they could have partnered with retailers on it. They have the resources. They have the systems. They have the capital. They have the need. There certainly is the desire. The added benefit would allow them do all the things I mentioned in my above post.

Lopetve: Tier-2 customers (folks actively avoiding you) are not actually customers. You concentrate on Tier-1 (ones who could be and want to be) and tier-3 (folks that don't know about your product somehow), while making sure Tier-2 doesn't do anything weird to you (consider Tier-2 the fanboys of the other side, in the normal marketing models) - https://cdn.slidemodel.com/wp-content/uploads/6539-02-blue-ocean-strategy-16x9-15-870x489.jpg gives a good example of the model normally used for this (this is partially my world). Again, this assumes that AMD's part can serve as a substitute - we don't know that yet, although you are correct, that many people will consider it a substitute. Then again, my RTX 2080TI is also a substitute - at least for now.

You are correct that customers actively avoiding you aren't customers. I had a typo in my post. I meant to say "the folks who can't afford them" instead of the "folks who don't want them".

It's an interesting model, but we're into business/sales philosophy again. I'm going to step back from that as it's ultimately something of a different topic.

....LOL....I'm getting tired typing so I'm going to rush the ending a bit.


Lopetve: Sale was made - to the distributor (scalper) or end user, the effect to the company bottom line is effectively nil. You're right that in the long term, nvidia DOES care about scalpers - because eventually they damage the brand, or can, but for a single launch? With everything else going on? Eh. Impact is minor right now - they're selling as many cards as they can make, a community of people has risen to try and GET the cards, many of us are just shrugging and paying the scalpers (if we can - I'm staring at a Gigabyte 3090 on ebay right now for a reasonable price), and there's still not a competitor. If they find a way of dealing with this and getting parts into peoples hands over the next 6 months, everyone will be happy and it will be mostly forgotten before long.


I do agree with you that the damage from this launch isn't horrid. It's the succeeding launches that will cause a problem. That's why they should be implementing some sort of reservation system now. If it can't be executed by the holidays then at least by the mid-cycle refresh.


Lopetve: Now, if Big Navi is truly competitive, and they can't figure out how to get yields up, or the scalpers find a way to totally dominate the market (profit margins there do matter to them, since they're effectively running on retail cost of goods sold), then they have to make a decision - Reebok/Nike/etc don't have issues with scalpers, as they've made the decision to accept that distribution model... Nvidia may not, as their products may not be unique enough for it to work or be valid (or the outcry might be significant enough to threaten the other lines of business), but until we see what their competition has, we just don't know.

I agree.

Lots of unknowns here. It's far more complex than "just throw up a preorder deposit system", which is the point I'm trying to make - because that's not as easy as it sounds. Same argument I get into with folks doing rental of enterprise datacenter equipment - if your customer's accounting isn't really built around operational expenses, and is built around capital instead with either EBITDA or capital amortization as key, then it doesn't matter how cost effective your solution is - they're not changing their accounting model to buy a bit of kit. The CFO would shoot you.

We've touched on a lot of issues ranging from philosophy to fiscal. It's been a good roundabout and I've enjoyed it. I won't be responding to the next one as these are getting lengthy, and we should have it over a beer instead. That said, I look forward to reading your response if you decide to write one.

I do agree that it is more complicated then throwing up a system. There are legal and financial issues to consider. However, I hope you agree that Nvidia has the resources to rapidly implement a system if they desire, and must make a choice on their future sales model.

Cheers!
 
Last edited:
Top