Nvidia has triumphantly exited markets such as chipsets or taken a step back from other markets such as Denver/x86. They seem to be setting the stage for yet another rearward advance. Which major market are the tea leaves prognosticating about this time around? Stick with us for a bit of back story before we get into that detail.
Before we get in to the meat of this, let’s do something to save us a lot of time and letters, going through some history. SemiAccurate and this author in particular has a long history of outing Nvidia stories long before anyone believes they could be real. With semiconductor design cycles running between 2-4 years, longer for some projects, decisions made today won’t be seen in public for years. Those with sources and ears to the ground can give an accurate picture of the goings on at a semiconductor company years ahead of time with very high accuracy.
Project cancellations are the easiest of the major changes to spot. Teams that should be working on the successor to project X being moved to a different line and no replacements named are an obvious sign. Roadmaps that end are another but there are many others, and how the company messages things are another good leading indicator. Nvidia in particular has a way of laying the groundwork for an impending rearward advance that is playing out as we speak.
Moving back to the history, let’s go over some of the highlights for SemiAcurate predictions on Nvidia. The author made a call that Nvidia had no chance of hitting the Q4/2012 or Q1/2013 dates it promised analysts at CES for Project Denver. Although it was directly stated to multiple analysts in one-on-one ‘private’ meetings, the company at the time knew there was no chance of it happening but directly said otherwise.
That said, SemiAccurate was the only one to say it and then described the core down to the pipeline level. For the record Denver is still unreleased for the reasons we said and a 2014 release is borderline at this point in time. The Tegra roadmap SemiAccurate published years ago was almost dead-on, later renamings aside. Compare and contrast those to the official messaging if you are really bored this holiday season. Suffice it to say that we know the Tegra roadmap quite well.
Then there are the GPUs, another topic SemiAccurate does not live up to its name on. From determining the root cause of the bump failures to calling the company out for shipping known defective parts to Apple, we are usually far ahead of the curve. Killing off parts preemtively, tapeouts, low yields, faking samples, roadmap slips, cancellations, personnel changes, and even big wins are all known to us. There are some mistakes too but we will stand by our track record of bringing you Nvidia news months or years ahead of the rest. If you are still bored after going over the Tegra roadmap history, feel free to do the same for GPUs.
Why are we going over this history? Because the latest one is a pretty big bang that won’t be seen for quite a while. If you go over the available data carefully, there is only one conclusion you can draw. Just like their exit from the chipset business, don’t expect them to admit it until it is both long dead and they can pin it on some other tangential external event to deflect blame.
Note: The following is analysis for professional level subscribers only.
Disclosures: Charlie Demerjian and Stone Arch Networking Services, Inc. have no consulting relationships, investment relationships, or hold any investment positions with any of the companies mentioned in this report.
Latest posts by Charlie Demerjian (see all)
- Globalfoundries 7nm process isn’t even close to the name - Sep 26, 2016
- ARM upgrades realtime offerings to v8-R and adds Cortex-R52 - Sep 21, 2016
- Everspin and Globalfoundries team up for embedded ST-MRAM - Sep 15, 2016
- Intel’s Xpoint is pretty much broken - Sep 12, 2016
- ARM adds 2048-bit vectors to v8A with SVE - Sep 7, 2016