Solved! Qualcomm buys Nuvia for $1.4 Billion

[Gideon]

Quite an absurd valuation for a startup without a single project yet.

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

Anyway, on top of Server chips Qualcomm should again have top-performing custom ARM cores for mobile SoCs in the future (judging by the development team, possibly better than Apple)

EDIT: and looks like laptops as well

NUVIA CPUs are expected to be integrated across Qualcomm Technologies' broad portfolio of products, powering flagship smartphones, next-generation laptops, and digital cockpits, as well as Advanced Driver Assistance Systems, extended reality and infrastructure networking solutions.

 

[beginner99]

Anyway, on top of Server chips Qualcomm should again have top-performing custom ARM cores for mobile SoCs in the future (judging by the development team, possibly better than Apple)

Assuming said team doesn't bail. But letting themselves get bought up already seems that was their plan all along or the offer they got was simply too good.

 

[moinmoin]

But letting themselves get bought up already seems that was their plan all along or the offer they got was simply too good.

That was likely their plan all along considering the current market circumstances (ARM designs becoming a force of their own, custom designs become rarer, smaller cutting edge ARM licensee shops shuttering left and right). A lot of startups are not built to reach self-sufficiency at some point but to be a nice addition to a bigger player's portfolio. With Qualcomm they essentially struck gold, there are not many other companies that have Qualcomm's market size and financial capability.

 

[Gideon]

I still can't get over the valuation.

1. They announced the company 15th November 2019
2. They sold it for $1.4 billion in January 2021 (essentially a year after the announcement), without a single product shipped (or close to shipping before 2022 most likely)

Essentially the entire valuation is based on their founders/engineers track-record (though I'm sure Qualcomm peeps have surely seen some simulations before this buy).

Then again, if your track-record looks like that:


Still, an awful lot of expectations weighting on their shoulders.

 

[Qwertilot]

Eh, its a far more rational valuation than a lot of the things that the stock market is doing right now!

 

[amrnuke]

I still can't get over the valuation.

1. They announced the company 15th November 2019
2. They sold it for $1.4 billion in January 2021 (essentially a year after the announcement), without a single product shipped (or close to shipping before 2022 most likely)

Essentially the entire valuation is based on their founders/engineers track-record (though I'm sure Qualcomm peeps have surely seen some simulations before this buy).

Then again, if your track-record looks like that:
View attachment 37693

Still, an awful lot of expectations weighting on their shoulders.

This is a good buy, IMO.
Regarding what they're paying, $1.4 billion seems like a lot but Qualcomm made $4.3 billion in profit in Q1-Q3 2020, so I'm not sure this is a huge expense considering their market stance competing with Apple in mobile phones, and the presence of Qualcomm chips in Chromebooks.

It's at least a little spicy to note that Apple and Qualcomm settled litigation in 2019 with Apple paying Qualcomm $4.5 billion for various patent infringements and royalty non-payments. Qualcomm then buying a company comprised of former Apple engineers less than 2 years later is quite interesting through that lens.

 

[moinmoin]

I still can't get over the valuation.

1. They announced the company 15th November 2019
2. They sold it for $1.4 billion in January 2021 (essentially a year after the announcement), without a single product shipped (or close to shipping before 2022 most likely)

Note that the first date is not necessarily when the company started working but when it went public due to completing its first funding round (where it raised $53m). Altogether it raised $293m in two rounds from 13 investors. I'm not that familiar with the startup market but the ratio (~4.78:1 sale price vs investment) intuitively seems to be on the lowish side to me.

 

[trexfromouterspace]

It's not a bad purchase at face value, but I'm highly skeptical because it's Qualcomm doing the buying. They've driven their cpu teams into the dirt over the last few years. Hopefully they don't repeat that with Nuvia.

 

[randomhero]

Wow! Good purchase! They bought top dogs in cpu design.

Unrelated: Tons of big news today!

 

[NTMBK]

Goodbye Nuvia servers, it was a glorious dream.

Nice to see Qualcomm getting back into custom CPU architectures, it's been boring without them.

 

[Gideon]

Goodbye Nuvia servers, it was a glorious dream.

Nice to see Qualcomm getting back into custom CPU architectures, it's been boring without them.

Are you certain this ends their server push? Why would qualcomm just give up a market now that they have the tools to compete ther

 

[HurleyBird]

If there's even a 50% chance it brings their custom cores within 10% of Apple's, it's a steal.

 

[Tup3x]

If there's even a 50% chance it brings their custom cores within 10% of Apple's, it's a steal.

...and if those would end up in Windows laptops things would get rather interesting.

 

[DisEnchantment]

I highly doubt a lot of people were surprised. This is payday.

However, if it means more performance for QC SoCs I am a bit excited about it. We are using a bunch of QC APQs and working with upcoming APQs as well and it is interesting what the future brings.
Although I am not sure I will be stuck in this same role by the time such products come to market. These things take a lot of time to materialize.

However, on the server side I still am bitter with my Centriq experience.

It remains to be seen how this is going to look like given that QC already gave up developing custom Cores.

 

[NTMBK]

Are you certain this ends their server push? Why would qualcomm just give up a market now that they have the tools to compete ther

The press release has zero mention of servers. It's all about Snapdragon.

 

[itsmydamnation]

You think Nuvia would have to be bought by someone, the big issue for them had to be patients,
I think its a good move by qualcomm, gives clear advantage over vanilla ARM SOC's. I wonder whats going to happen with big little for snapdragon, keep using ancient A55 or will Nuvia do both?

 

[Gideon]

You think Nuvia would have to be bought by someone, the big issue for them had to be patients,
I think its a good move by qualcomm, gives clear advantage over vanilla ARM SOC's. I wonder whats going to happen with big little for snapdragon, keep using ancient A55 or will Nuvia do both?

This was discussed on twitter. AFAIK they really can't use A55 with custom cores well (at least not in the same cluster using shared higher-level caches):

https://twitter.com/x/status/1349371657606074373

 

[Gideon]

The press release has zero mention of servers. It's all about Snapdragon.

Ian Cutress discusses it on his channel:

He finds it unlikely as part of the reason Nuvia was founded in the first place was supposedly that founders were tired of making smartphone SoC's and Apple wasn't really interested in servers. Why would they sell their company to anyone that would bar them from continuing with that work and force them excluseively to mobile SoC's yet again. IMO it would be a big loss to the ARM server world if these ambitions get canned.

EDIT:
And a tweet

https://twitter.com/x/status/1349465224638836744

 

[beginner99]

I still can't get over the valuation.

1. They announced the company 15th November 2019
2. They sold it for $1.4 billion in January 2021 (essentially a year after the announcement), without a single product shipped (or close to shipping before 2022 most likely)

Large part of the valuation might the bo avoid a future much higher price when there is competition bidding.

Eh, its a far more rational valuation than a lot of the things that the stock market is doing right now!

Like Tesla. Or anything else really (but it's limited mostly to US)

 

[coercitiv]

He finds it unlikely as part of the reason Nuvia was founded in the first place was supposedly that founders were tired of making smartphone SoC's and Apple wasn't really interested in servers. Why would they sell their company to anyone that would bar them from continuing with that work and force them exclusively to mobile SoC's yet again. IMO it would be a big loss to the ARM server world if these ambitions get canned.

And yet Nuvia themselves compared projected core performance using competing mainstream cores and an arguably consumer oriented workload (GB5). For a company focused on server products they went out of their way to signal their IP was great for consumer applications.

Look at this, not one server oriented core in sight:

Spoiler: Geekbench 5 Performance per watt

Sure they later argued there was correlation to be observed between SPEC and GB scores, but in the light of the latest news one would have to ignore a lot of signs to keep thinking about server products. IMHO we'll see consumer products based on their IP before anyhting else.

 

[NTMBK]

Ian Cutress discusses it on his channel:

He finds it unlikely as part of the reason Nuvia was founded in the first place was supposedly that founders were tired of making smartphone SoC's and Apple wasn't really interested in servers. Why would they sell their company to anyone that would bar them from continuing with that work and force them excluseively to mobile SoC's yet again. IMO it would be a big loss to the ARM server world if these ambitions get canned.

EDIT:
And a tweet

https://twitter.com/x/status/1349465224638836744

Or Nuvia rapidly found out that the ARM server market just isn't suitable for merchant silicon vendors, and none of the big cloud providers wanted to purchase them.

 

[Gideon]

Or Nuvia rapidly found out that the ARM server market just isn't suitable for merchant silicon vendors, and none of the big cloud providers wanted to purchase them.

Might indeed be the case. Though I still think they would have licensed the cores, if they were a big enough jump from ARM defaults.
I do agree though that with Amazon, Google and MIcrosoft pursuing their own chips, buying off-the-shelf probably would not have worked.

 

[NTMBK]

Might indeed be the case. Though I still think they would have licensed the cores, if they were a big enough jump from ARM defaults.
I do agree though that with Amazon, Google and MIcrosoft pursuing their own chips, buying off-the-shelf probably would not have worked.

Genuine question- do ARM license terms allow for 3rd parties to license core designs? I was under the impression that it was not permitted, and that an architecture license only allows you to build custom cores for your own chips.

 

[scannall]

Genuine question- do ARM license terms allow for 3rd parties to license core designs? I was under the impression that it was not permitted, and that an architecture license only allows you to build custom cores for your own chips.

You're exactly right on that.

As far as Nuvia goes, it's been apparent to many of us it was just producing hype and generating IP for a buyout. Their business model wasn't strong enough for anything else.

 

[Doug S]

He finds it unlikely as part of the reason Nuvia was founded in the first place was supposedly that founders were tired of making smartphone SoC's and Apple wasn't really interested in servers. Why would they sell their company to anyone that would bar them from continuing with that work and force them excluseively to mobile SoC's yet again. IMO it would be a big loss to the ARM server world if these ambitions get canned.

I can think of 1.3 billion reasons why they would sell even if Qualcomm doesn't want to use it for servers. There will be a lockup period, but it will be limited and then they are free to leave and do another ARM server CPU startup if they want. Best of all, they'll have enough money they won't need hit up VCs for round one financing so they really will be free to do whatever they want and will be able to maintain majority ownership.

The timeline will be interesting here. Since Qualcomm's SoC for next year has to tape out in the next couple months (if it hasn't already) the earliest a Qualcomm SoC with a Nuvia core could appear would be two years from now to go into 2023 phones (and thus competing against Apple's A16 made on TSMC's 3nm) They will need to design their own small core, since Nuvia obviously didn't design one for their servers and as mentioned above integrating with an A55 would not work well. There may be some rework of their big core design if Qualcomm plans to have that made at Samsung, since Nuvia was likely targeting TSMC, as well as potentially removing some server type features (i.e. if it integrated wide SVE2 that'll need to be chopped out)

That's a lot to get done in 12 months, and won't leave a lot of time for working on the 2nd gen Nuvia core - which has undoubtedly been started but how far along it may be will depend on when exactly they thought they'd start fabbing the 1st gen one. If GW III and company don't plan to stick around, Qualcomm will at best get two SoC generations out of them, then the ones who don't follow them will have to run with it after that.