As per a Wall Street Journal article, NVIDIA is postponing data center GPU orders for customers who consider other hardware options. Groq, a new AI chipmaker whose advanced TSPs we’ve covered before, disclosed this information.
These customers are wary of exploring alternative AI accelerated hardware due to concerns that NVIDIA, being the undisputed market leader in AI acceleration, might retaliate by delaying their already placed orders.
Despite NVIDIA’s claims of fair supply practices and providing alternative access to compute hardware during shipment delays, Groq’s CEO, Johnathan Ross, asserts that certain customers are deliberately avoiding meetings with competing AI chipmakers.
These customers, who typically pay upfront, might endure delays of up to a year for new hardware shipments. Despite these delays, some customers opt to await updates from NVIDIA rather than exploring alternative options.
In early 2018, NVIDIA introduced the “GeForce Partnership Program” (GPP) in an attempt to foster stronger relationships between NVIDIA and its AIB partners for promoting the NVIDIA GeForce brand and technologies. However, the program faced significant backlash as AIB vendors were essentially compelled to promote aftermarket hardware exclusively under the NVIDIA brand.
ASUS, in particular, received criticism for rebranding all AMD hardware under the ROG (Republic of Gamers) branding to a new “AREZ” moniker as a result of this initiative, NVIDIA essentially imposed this initiative on its partners, offering benefits like increased shipments of graphics cards allocated to those vendors or possibly a lower unit price in exchange.
NVIDIA has gained a reputation for burning bridges, often making last-minute changes to its product lineup and pricing. This practice led to EVGA’s departure from the graphics card market altogether, as they incurred financial losses due to NVIDIA’s behavior, despite being an exclusive partner for well over a decade. Despite being one of the best-selling AIB vendors in North America, EVGA found it unsustainable to continue operating under these conditions and eventually quit altogether.
Almost fifteen years ago, BFG Tech, a respected US-based aftermarket manufacturer known for its lifetime warranties, faced a situation similar to EVGA’s recent challenges. In 2010, BFG decided to exit the GPU market entirely, despite being an exclusive aftermarket manufacturer for NVIDIA.
XFX, another AIB vendor, faced rejection from NVIDIA after exclusively producing NVIDIA graphics cards. When XFX sought to broaden its horizons by venturing into AMD/ATI aftermarket graphics cards, NVIDIA severed ties with them as an official partner.
Like EVGA, BFG cited similar reasons: the margins associated with the sale of GeForce hardware were insufficient. They were being undercut in the market by other AIB vendors who had access to more units of more cost-effective offerings.
It’s noteworthy that EVGA exited the GPU market around the peak and eventual decline of the Ethereum mining boom. There’s a circulating theory that EVGA placed a large quantity order of NVIDIA graphics cards during this period when supplies were scarce and scalping was rampant.
NVIDIA, capitalizing on the high demand, raised prices for chip sales to AIB vendors. However, with the end of Ethereum mining, GPU availability became abundant, leading to a rapid drop in prices. It’s believed that EVGA was left with a substantial bill, having paid an excessive amount for graphics chips that were now severely undervalued. Consequently, they incurred significant losses on every sale.
The GeForce Partnership Program, or what I like to dub the GeForce Propaganda Program, essentially coerced AIB vendors into sidelining AMD Radeon hardware by leveraging recognizable branding, such as ASUS and their ROG products exclusively to GeForce. Failure to comply could result in penalties like decreased GPU allocations.
Although margins for NVIDIA graphics cards are lower compared to AMD Radeon offerings for AIB vendors, companies like MSI, Gigabyte, and ASUS manage to thrive thanks to the sheer volume of units sold, rather than prioritizing higher margins on fewer units sold.
With NVIDIA now producing and distributing their own versions of GeForce GPUs called “Founder’s Edition,” they are effectively sidelining AIB vendors and directly pocketing the profits. Previously, NVIDIA would manufacture a “reference” board design, which AIB manufacturers would then purchase and distribute under their branding. However, with the introduction of Founder’s Edition cards, NVIDIA is bypassing this process and capturing revenue directly.
NVIDIA made the decision to cancel the GeForce Partner Program shortly after its announcement. The company clarified that the program had been misunderstood and emphasized that it was never meant to restrict the availability of AMD products or stifle competition. However, it’s worth noting that ASUS swiftly abandoned the idea of releasing an AMD Radeon-based ROG product, a move that attracted significant attention given ROG’s prominence in the DIY PC market.
Essentially, NVIDIA, as the market leader, has been employing questionable tactics for a significant period. This includes pushing their proprietary “GameWorks” SDK onto developers, hindering the adoption of AMD’s innovative open-source technologies, the recent incorporation of “Ray Tracing” implementations that compromise performance for minimal visual gains with GeForce RTX. Additionally, the normalization of upscaling, even on enthusiast-grade hardware, has regressed the industry as a whole.
Former Senior Vice President and General Manager at AMD Radeon, Scott Herkelman, indicates that such occurrences are more widespread than perceived. He asserts that NVIDIA has been employing comparable strategies with data center clients, OEMs, AIBs, affecting resellers, and even influencing the press.
For instance, TechPowerUp once claimed that certain YouTuber hardware reviewers and journalists were subtly promoting a specific brand by displaying NVIDIA GeForce RTX/GTX graphics card boxes in the background of their videos.
However, TechPowerUp removed the article shortly after publishing it due to backlash, particularly from individuals like Steve from GamersNexus. Steve strongly objected to being linked with subliminal promotion and even threatened action against TechPowerUp.
I can’t help but ponder how these review channels sustain themselves financially. Acquiring the latest products in the DIY PC space is a significant investment, and while independent reviews are meant to be just that, some channels, like GamersNexus, appear to be commercialized.
It’s unlikely they generate sufficient revenue solely from selling merchandise like Chinese made mugs with their logo on them.
Herkelman additionally pointed out that, in the wake of the controversy surrounding the GeForce Partner Program (GPP), NVIDIA has taken a more cautious stance by refraining from formal written agreements. The company encountered significant scrutiny from both the media and graphics enthusiasts when information about the GPP came to light, ultimately resulting in its abandonment.
Throughout their history, NVIDIA has consistently employed underhanded tactics to maintain control. They have secured a dominant position in the AI/Deep Learning industry through questionable means, including the monopolization of CUDA and donating advanced hardware to numerous universities for years.
Despite not always offering the most advanced AI hardware, NVIDIA’s dominance persists due to widespread industry support for GeForce hardware simply because their monopoly has been entrenched within the industry from the very beginning, so it wouldn’t surprise me if NVIDIA were to give customers the cold shoulder for shopping around towards some of the other corporations delving into Artificial Intelligence hardware, such as AMD’s MI300 or others.