J.P. Morgan said in its latest Asia-Pacific equity industry survey that Nvidia’s Rubin Ultra-generation NVL576 switch tray is reportedly set to adopt PTFE, or polytetrafluoroethylene. The bank added that this does not overturn the upgrade roadmap for high-speed copper-clad laminates, or CCL, and should be viewed as one part of a broader competition among materials.
PTFE in NVL576 does not cancel the high-speed CCL roadmap
Taiwan stocks surged last week on the Rubin Ultra “glass-less” theme, with PTFE-related names such as Taiflex and Asia Electronic hitting limit-up moves in succession as the market quickly priced in the possibility that PTFE could replace current high-speed CCL materials.
J.P. Morgan took a more restrained view in its report. It said PTFE adoption in the NVL576 switch tray is a real development, but not one large enough to invalidate the broader CCL upgrade roadmap.
The report framed the issue around Nvidia’s electrical performance thresholds. According to J.P. Morgan, Nvidia is testing a wide range of materials. PTFE is among the options because of its strong low-loss profile, but candidate status is not the same as final selection. The bank added that most M10 samples submitted by suppliers still sit one step away from the target, which leaves technical risk unresolved.
Hybrid designs are seen as the most practical route
J.P. Morgan said the industry’s technology path has shifted toward hybrid solutions rather than a single-material design. The approach it sees as closest to mass production pairs PTFE for low-loss electrical performance with glass-less M10 for processability.
The report gave two reasons a pure PTFE structure is not ready for direct mass production:
- Pure PTFE stack-up designs lack the process capability needed for high-temperature lamination.
- Drilling pure PTFE is difficult, and blending in M10 helps ease that problem.
J.P. Morgan said Nvidia’s target is to meet an electrical performance standard, not to commit to a specific resin brand. The only sample currently within 5% of the target is a laminate-only version with the glass fiber removed. Its loss performance is strong, but the board becomes softer and lacks structural strength, which makes large switch boards hard to manufacture at scale.
Glass-less boards still face engineering hurdles in switch applications
In theory, removing glass fiber can deliver better electrical performance. That is why the Rubin Ultra generation sparked the recent “glass-less” narrative. In practice, J.P. Morgan said switch boards face more complicated engineering constraints than flexible boards and substrate boards.
Fully glass-less mass-production cases already exist in flexible boards and substrate boards because those products can be reinforced with added components. Switch boards are much larger and do not lend themselves easily to extra structural support. Once glass fiber is removed, suppliers may run into weaker metal adhesion, open questions around high-voltage tolerance, and the difficulty of drilling softer materials. J.P. Morgan said it is still too early to conclude whether glass-less designs can overcome all of those issues.
For NVL576, the bank’s more constructive case is a hybrid split: glass-less M10 handles loss performance, while PTFE with glass fiber provides structural support. It described that division of labor as the most realistic compromise available now.
Rubin Ultra may serve as the first test point, with broader PTFE left for Feynman
J.P. Morgan also laid out a timeline for possible adoption. It said an M10/PTFE hybrid is most likely to appear first in the switch tray position of Rubin Ultra NVL576 8-rack clusters, and only there rather than across every board in the system.
For the single-rack Rubin Ultra configuration, the bank expects M8 to remain in place, implying a much smaller upgrade step.
A more aggressive all-PTFE design, the report said, may have to wait for the Feynman generation. Even then, that depends on whether drilling technology and PTFE lamination capacity issues are solved. If that happens, adoption could extend to areas such as multi-cluster NVLink switch boards.
Uncertainty may keep related stocks volatile for years
J.P. Morgan concluded that PTFE does not negate the high-speed CCL upgrade path and that the demand logic for M10 still stands. What remains unresolved is the complexity of the material mix and the difficulty of delivering stable yields.
The bank said shifts in technical direction are likely to keep moving market sentiment, forcing repeated swings in the shares of CCL makers, PCB manufacturers, and upstream materials suppliers.
For investors, the current price action in PTFE-linked names reflects expectations of a supply-chain reshuffle after successful qualification, rather than confirmed orders or secured capacity. That cautious reading is also the core message of J.P. Morgan’s report.