Should I hold off on an Agilex 5 purchase now that Agilex 2X is out?

I've been through a few of these generation handoffs, both as a buyer and sitting on the vendor side at a previous job, and the pattern is pretty consistent. The big flagship part gets announced first with a datasheet because it's what makes headlines, but actual orderable volume on that top part is tiny for the first couple of quarters. The mid-range parts, wich is where your money usually goes, show up 12 to 18 months after the announcement, sometimes longer. And even when the orderable part number exists, early allocation goes to the big accounts first. If your lead times are already painful, jumping to a brand new family makes that worse, not better, because new silicon always has allocation problems in year one. On the tools: yes, plan on them being rough. First releases of a toolchain supporting new silicon tend to have fit and timing quirks, incomplete IP, and bugs that don't get cleaned up until a couple of point releases. If you're spinning up a fresh project with a hard deadline, doing it on a mature toolchain with a part that's been shipping for a year or more is a much calmer experience. Early adopters on new families are basically unpaid beta testers, and that's fine if your schedule has slack. Yours doesn't sound like it does.

The other thing people forget is pricing. New generation parts launch at a premium and the previous generation gets cheaper as the vendor pushes the new stuff. If your design fits comfortably in an Agilex 5, you're likely to get a better deal on it now than you would have a year ago, and you'll actually have boards in hand.

My rule after getting burned once: never delay a real project for an announced part unless that part has a specific feature you cannot get anywhere else. A name and a slide is not a product. If the new family turns out great, your next project can use it, and by then the tools will actually work. Ask your manager the simple version of the question: does the project need anything the Agilex 5 can't do? If the answer is no, order it and stop worrying about it.

I made the opposite mistake a few years back, held off ordering for the better part of a quarter because a newer generation was on the slide deck, and then the launch slipped twice before I gave up and went with the older part anyway, which by that point had been shipping long enough that errata were actually documented somewhere findable. The delay cost me about four months of schedule for zero benefit. One thing I'd add to what's already been said, check the announcement carefully for whether the parts you'd actually buy are even on the roadmap slide or just the family name. Vendors compress a lot of vagueness into a single logo with a bunch of arrows under it.

We picked the newer family on a project two years back because the numbers looked tempting on paper. Six months later we were still waiting on boards while the old part I'd originally specced was sitting in stock at three distributors. Worse, the eval board for the new silicon cost way more and had half the community reference designs missing. This time around on a small side build I just went with the mature part and it was done in a weekend. One thing I'd check before you order though: how many board vendors support the Agilex 5 right now, since that saved me a pile of layout work.

Reading this thread as someone who just messes around with these boards at home on weekends, so take my take with a grain of salt. What I keep wondering is whether the older parts stay easy to buy for long, or if once the shiny new family lands everyone switches and the old ones slowly vanish from the normal shops. That whould kinda force the decision anyway. Also, if I bought the older one now and my little project grew, would I be stuck redoing everythign to move it to the newer chip later, or do the designs usually carry over?

Update: following the point about the old part sitting in stock at distributors, I actually ran availability checks this morning and the board I'd specced shows solid quantities at two distributors, so I put the quote in. One thing I keep wondering though, for the person whose launch slipped twice: did your rep give you any honest signal upfront, or was it all press release optimism untill the delays hit?