(Anecdote) - Mobile x86 / memory supply for corporate laptops

[Xebec]

A friend of mine is rebidding a corporate laptop purchase agreement for a company, dealing with 3 major OEMs. The pricing and patterns of responses by the OEMs are a bit telling on the state of the x86 mobile supply chain:

- All 3 OEMs are discouraging AMD - both verbally, and through high pricing (+20-40% premium). 2 of these OEMs are traditionally AMD friendly.
- exception: One lower end SKU is priced competitively with Lunar Lake, but is very clearly a lower tier (performance) product in a corporate laptop environment
- I haven't personally been involved in corporate bids for a few years, but traditionally AMD is hungry for market share in laptops and will offer strong discounts

- They are all steering the bid from Arrow Lake-H to Lunar Lake (-V) :
- All 3 OEMs raised prices significantly on ARL-H relative to where they were bid a year ago. "Prices through the roof"
- One OEM indicated Arrow Lake availability of ~ 1 Quarter away vs. 2-3 weeks for Lunar Lake
- Lunar Lake pricing is "not expensive" compared to the other two (i.e. hundreds $ less), even in cases where it has a bigger battery than ARL-H or AMD

- "No Panther Lake supply or announcement until ~ May" from one OEM. The other OEMs are also not proposing Panther Lake.


.. Reading into this, I think:
1. Wafers that can produce Ryzen (mobile) are likely being prioritized elsewhere by AMD. (AMD is traditionally hungry to grow laptop market share)
2. Arrow Lake-H is in short supply compared to Lunar Lake. (High cost and low availability)
3. Lunar Lake is both in supply, and likely significantly cheaper due to having RAM bundled from Intel.
4. Panther Lake is definitely not a high volume at launch scenario. (I think we all knew this already..)

......

*A few other notes: 1. There are no TAA requirements for these devices (i.e. they can be sourced from anywhere), 2. The bid quantity is for a large company, but not a huge company (i.e. not 100K+ bid size). 3. Lunar Lake laptops tend to save some manufacturing costs vs. ARL-H and AMD Ryzen Pro laptops by being thinner/lighter (less material) due to lower power envelope.

 

[MKWVentures]

A friend of mine is rebidding a corporate laptop purchase agreement for a company, dealing with 3 major OEMs. The pricing and patterns of responses by the OEMs are a bit telling on the state of the x86 mobile supply chain:

- All 3 OEMs are discouraging AMD - both verbally, and through high pricing (+20-40% premium). 2 of these OEMs are traditionally AMD friendly.
- exception: One lower end SKU is priced competitively with Lunar Lake, but is very clearly a lower tier (performance) product in a corporate laptop environment
- I haven't personally been involved in corporate bids for a few years, but traditionally AMD is hungry for market share in laptops and will offer strong discounts

- They are all steering the bid from Arrow Lake-H to Lunar Lake (-V) :
- All 3 OEMs raised prices significantly on ARL-H relative to where they were bid a year ago. "Prices through the roof"
- One OEM indicated Arrow Lake availability of ~ 1 Quarter away vs. 2-3 weeks for Lunar Lake
- Lunar Lake pricing is "not expensive" compared to the other two (i.e. hundreds $ less), even in cases where it has a bigger battery than ARL-H or AMD

- "No Panther Lake supply or announcement until ~ May" from one OEM. The other OEMs are also not proposing Panther Lake.


.. Reading into this, I think:
1. Wafers that can produce Ryzen (mobile) are likely being prioritized elsewhere by AMD. (AMD is traditionally hungry to grow laptop market share)
2. Arrow Lake-H is in short supply compared to Lunar Lake. (High cost and low availability)
3. Lunar Lake is both in supply, and likely significantly cheaper due to having RAM bundled from Intel.
4. Panther Lake is definitely not a high volume at launch scenario. (I think we all knew this already..)

......

*A few other notes: 1. There are no TAA requirements for these devices (i.e. they can be sourced from anywhere), 2. The bid quantity is for a large company, but not a huge company (i.e. not 100K+ bid size). 3. Lunar Lake laptops tend to save some manufacturing costs vs. ARL-H and AMD Ryzen Pro laptops by being thinner/lighter (less material) due to lower power envelope.

great input.

One model, I am working on is : What is the response to these incredible DRAM price increases? OEMs are going to have to figure it out. I am not sure but prioritizing lunar lake makes sense.

This is for corporate, but for overall, : just re-checked Well over half of all PC CPUs sold by Intel are Intel 7 or older. Similar for AMD. MOST PC CPUs sold were launched before 2024. I knew this was true for DC CPUs (slow DC ramps). I did not realize how bad it was in PC CPUs until I checked.

PC OEMS should be asking which CPU runs best on minimal DRAM

Panther lake will not have a positive impact on Intel in 2026.

 

[blueone]

I wonder if Lunar Lake is cheaper because they were built last year when DRAM was priced lower, and Intel is selling these CPUs out of inventory. Also, I would bet Lunar Lake motherboards are cheaper too (no SO-DIMM slots, simpler routing so fewer layers required).

 

[hist78]

"No Panther Lake supply or announcement until ~ May" from one OEM. The other OEMs are also not proposing Panther Lake.

"Strangely, the first Intel BSPD (PowerVIa) product, Panther Lake, is for laptop computers. How do Intel and the OEMs overcome the cooling and battery drain problems? There isn’t a practical or inexpensive way to use liquid cooling in a laptop PC, unless they throttle down the performance to control heat and extend battery life. How does Intel solve these problems and this dilemma?

Why didn’t Intel apply BSPD to a processor for servers or desktop PCs first, where liquid cooling is feasible and meaningful?"

Post in thread 'Intel’s 18A rumors meet a thermal brick wall says SemiWiki'

Today at 12:04 PM

All of which is are the underlying reasons behind why TSMC recommends A16 (with BSPD) only for "HPC-type devices -- dense power grid with active cooling". For these BSPD definitely has more pros (including better PPD) than cons, which is why it's being adopted for such devices -- also higher die cost doesn't really matter if the selling price is many thousands of dollars each. Since this (CPUs) is also Intel's main market, going BSPD-only works for them, it's the right choice.

For more "normal" chips -- most of the market by number of chips, and probably also wafer volume...

• hist78

 

[Xebec]

"Strangely, the first Intel BSPD (PowerVIa) product, Panther Lake, is for laptop computers. How do Intel and the OEMs overcome the cooling and battery drain problems? There isn’t a practical or inexpensive way to use liquid cooling in a laptop PC, unless they throttle down the performance to control heat and extend battery life. How does Intel solve these problems and this dilemma?

Why didn’t Intel apply BSPD to a processor for servers or desktop PCs first, where liquid cooling is feasible and meaningful?"

Post in thread 'Intel’s 18A rumors meet a thermal brick wall says SemiWiki'

Today at 12:04 PM

All of which is are the underlying reasons behind why TSMC recommends A16 (with BSPD) only for "HPC-type devices -- dense power grid with active cooling". For these BSPD definitely has more pros (including better PPD) than cons, which is why it's being adopted for such devices -- also higher die cost doesn't really matter if the selling price is many thousands of dollars each. Since this (CPUs) is also Intel's main market, going BSPD-only works for them, it's the right choice.

For more "normal" chips -- most of the market by number of chips, and probably also wafer volume...

• hist78

It's probably complicated, but a few guesses/thoughts:

- It's a brand new node, Mobile is usually first production on a node because of yields (mobile has smaller dies), and lower clock speeds on early versions of the process (mobile less demanding than desktop). Panther Lake tops out at 5.1 GHz which can be achieved at reasonable voltage even on old Intel 7.

- GAAFET provides a good reduction in voltage per frequency, further reducing power draw (i.e. best benefit to mobile and server, desktop secondary)

- BSPD also has the benefit of lower power usage (thermal leakage differences aside) at ISO frequency, at least according to Intel's slide. Worth noting Intel did test BSPD on an earlier node internally - so they should have some real data to support this.

That leaves only the 'thermal sandwich' problem which makes the chip harder to cool. In this situation, I think Mobile is still the best place to start with GAAFET+BSPD again because the total power envelope is lowest. A desktop/server sized cooler doesn't necessarily mean it can get the heat out of the chip any faster - they still have to deal with the resistance of heat transfer at the chip level. (It'll be interesting to see teardowns about how thin or thick the materials are on/around the CPU tile)

Some info: https://spectrum.ieee.org/backside-power-delivery

To take some of the risk out of this high-wire act, the company has built and tested processor cores composed of Intel’s current generation of transistors combined with PowerVia. The resulting cores saw more than a 6 percent frequency boost as well as more compact designs and 30 percent less power loss. Just as important, the tests proved that including backside power doesn’t make the chips more costly, less reliable, or more difficult to test for defects.