Nvidia Just Invited a Rival Chipmaker Into Its Own Racks. Here’s Why
Nvidia is letting a rival chipmaker plug directly into its own racks, and the move only looks like a concession until you follow the money through every layer of the data center.
A company that sells the most sought-after AI accelerators on earth is now helping a competitor’s accelerator slot into its own racks. NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) has opened part of its infrastructure ecosystem to d-Matrix, a Microsoft (NASDAQ:MSFT)-backed inference chip startup last valued at approximately $2 billion.
D-Matrix plans to wire its Raptor processors into Nvidia-powered systems using the NVLink Fusion interconnect, with the first rack-scale products expected in 2027. That would be after Raptor tapes out near the end of 2026.
Nvidia has not disclosed the financial terms of the arrangement. The counterintuitive idea worth exploring is whether Nvidia earns more as the toll collector for AI data centers than it would by trying to win every chip sale inside them.
What NVLink Fusion Actually Does
NVLink is the high-speed fabric that lets racks of GPUs behave like one computer. NVLink Fusion extends that fabric to selected partners so their silicon can plug into Nvidia’s rack, share memory bandwidth, and speak the same language as Nvidia’s GPUs and CPUs.
For d-Matrix, that connection matters more than the chip itself. Inference accelerators live or die on how quickly they can reach model weights and pass tokens back to a host system.
For Nvidia, licensing the fabric turns a proprietary bus into an industry standard. Marvell (NASDAQ:MRVL) joined NVLink Fusion earlier this year, and Groq is being integrated at the rack level, with the CFO calling out a “Strategic partnership with Marvell via NVLink Fusion” in the most recent quarter.
Chief executive Jensen Huang framed the logic bluntly on the last earnings call: “AI has reached its inflection point. It’s doing useful work. Its tokens are productive and profitable. Now, compute is revenue.”
Toll Collector Case
Even when a partner supplies the principal accelerator, Nvidia keeps collecting from networking, CPUs, NVLink switches, BlueField storage, and CUDA software. Data-center networking alone brought in $40.31 billion, up 138% year over year last quarter.
Rack economics reinforce the point. Nvidia now pegs data-center revenue per gigawatt at roughly $18 billion for Hopper, $25 billion for Blackwell, and $40 billion for Vera Rubin, because the AI factory now includes far more than GPUs.
Accelerator leadership changes hands faster than interconnect standards do. Once hyperscalers standardize their racks on NVLink Fusion, swapping GPUs for a rival part still leaves Nvidia inside the machine, charging for the plumbing. Someone has to power, cool, and network all that buildout, and we pulled together seven of those non-chipmaker suppliers in a free AI infrastructure report.
Total revenue reached $96.22 billion, up 105.8% year over year, and management guided the current quarter to $108.0 billion. Ecosystem expansion is funding that curve.
Where This Could Go Wrong
Opening the racks hands specialized inference competitors a credible on-ramp to a workload where Nvidia’s GPU margins are fattest. Inference is precisely where custom silicon competes best on cost and power.
If Raptor and similar parts prove several times more efficient per token, hyperscalers could shift the highest-volume workloads off Nvidia GPUs while keeping the fabric. Nvidia would still collect a toll, but a smaller one.
Gross margin already tells part of that story. Management expects margins to bottom in the 71% to 72% range in the fourth quarter before recovering.
The hindsight-mistake scenario is straightforward: NVLink Fusion becomes the standard, third-party accelerators capture inference share, and Nvidia’s per-rack dollar content stops climbing. That risk is real.
Is NVDA Stock a Buy?
At $218.29, Nvidia trades at a forward P/E near 24x with analyst consensus at $327.18.
Ecosystem partnerships with d-Matrix, Groq, and Marvell strengthen Nvidia’s position more than they erode it, because the fabric is the durable asset. Call it a Buy, though I’d be more careful over the near term, as rising oil prices and Treasury yields could spell trouble for the broader market as a whole, and NVDA stock can fall even if earnings do well.
Contact [email protected] for any questions or corrections.
Omor Ibne Ehsan is a writer at 24/7 Wall St. He is a self-taught investor with a focus on growth, cyclical, and dividend equities that have strong fundamentals, value, and long-term potential. He also has an interest in high-risk, high-reward investments such as penny stocks.
