Thursday, December 7, 2023

Intel Foundry External Wafer Sales Forecast

 

Intel Foundry Forecast Dec 2023 (Mark Webb, MKW Ventures Consulting, www.mkwventures.com)

We have looked at Intel Plans, Customer reactions, and competitor reactions to develop a forecast for Intel Foundry

Today, Intel IFS  includes sales of mask making products and services in its foundry revenue. It appears this is actually the largest revenue segment in IFS per Intel 10Q reports. Intel also includes packaging and assembly services for external companies.

Intel actual wafer sales to other companies (wafer foundry) is quite small today. We estimate $100-200M per quarter. This would put them as the 15-20th largest wafer sales foundry in the world.

 

In the future, Intel has plans for technologies and is sampling with many companies. If they execute, they will have leading edge technologies and capacity to ramp them.

We expect all major customers to move slowly with small portion of foundry allocation to Intel until 2027 timeframe. Many are planning to use Intel as a backup or to provide price negotiation pressure. TSMC is a dominant force in foundry and this is not changing. TSMC considers Intel a large customer and not a threat in foundry.  They may or may not be correct.

In our model, Intel always spends 2x the dollars outsourcing to TSMC as they receive in wafer foundry revenue from others.

The reported accounting will all change next year. Intel will try to report fab business as independent business selling to Intel BUs. They will quickly become a top 3 foundry. They will adjust the allocated cost to achieve the goals they want (ie: Allocate low cost to BUs to compare for TSMC foundry work.

The forecast for Intel External wafer sales is below: Note that this is based on Intel plans. If they do not execute or gain share, then they will go to the “Plan B” (Call us for what Plan B is)



We are at IEDM 2023 Dec 10-13 to discuss how we update this model each month. Call or text to set up meeting.

Mark Webb





Wednesday, December 6, 2023

Memory Market Update Dec 2023 Edition

 As we predicted in Feb, March, May, August and October, the memory market is recovering in Q4 2023. 

Reminder, we predicted this before it even bottomed and we are using the same cycle graph. 

The reason is that the market always behaves in cycles and the only real movement is inventory and price... you can see all the details on our website and 2023 FMS presentation which quantifies causes and effects. www.mkwventures.com

Summary of downturn: the Total bits sold in 2020-23 is Exactly what we expected in 2019... there was no drop. Customers just bought them all in 2020-early 2022. For some reason, no one noticed inventory build at customers that was clear (See earnings report transcripts for Q2 2022).

So where are we now?:

  • Memory prices are recovering
  • Since they are recovering, people are incentivized to buy and build inventory now.
  • the recovery will get sharp until we reach profitability
  • It is not clear how profitable we will get. That depends on inventory shortages in 2024/5 which we track.
  • If we fall into the trap that [enter hype technology or geopolitical event] is changing long term demand, the 2026 crash will be bad as well. 
  • DRAM bits will grow 15-18%. NAND Bits will grow 20-25% long term. Beware when people say "DRAM bits grew 15% in Q4 2024, We don't see any evidence of double ordering" 
  • Obviously there will be a downturn in 2026. We help you predict whether it will be Minor (20% revenue drop) or Major (50% revenue drop)
  • We can advise memory companies on how to meet demand without getting played by the customers and causing oversupply
  • Call or text us to discuss more and learn how you can d




Mark Webb




Tuesday, December 5, 2023

HBM Market Forecast and Current Facts

 


HBM Revenue Model Key Facts:

  • HBM Bits are <5% of DRAM Bits Today
  • HBM Revenue is <10% of DRAM Revenue Today
  • HBM Price/bit is >5x  of DDR4/DDR5 DRAM
  • HBM Cost/bit is >3x of DDR4/DDR5 DRAM
  • Today's Market is supply constrained, sole source per project, little price competition
  • Tomorrow's will be price competition, aggressive moves to obtain market share

Call or text to set up meeting to discuss how the market will change over time, who will win, and why each of the three major DRAM suppliers is in a completely different starting position 

We are at IEDM in San Francisco Dec 10-13. Call or text to meet in person or set up a zoom call.


Mark Webb



Monday, December 4, 2023

Mark Webb, MKW Ventures Consulting at IEDM Dec 10-13 2023.

  



7 Topics for Discussion at IEDM 2023. Mark Webb, MKW Ventures Consulting LLC

              


                  1 HBM memory: It is the fastest memory and it powers the AI compute industry.

a.      How much does it really cost and how much capacity is needed and what will the supply chain be in 3 years?

b.      Why is only 10% of the DRAM memory in Nvidia Servers actually HBM.

c.      How do we think about “super near”, “really near”, ”sorta near”, “pretty far”, “very far” memory in Servers today and in the future? 2LM is now 5LM. Do emerging memories change this?

2 The Memory market is recovering exactly as we predicted in March 2023 (check our presentations and predictions). We also gave details to the memory market cycle at FMS 2023 and how it has nothing to do with Pandemics, recessions, AI, inflation, supply chain issues. It has to do with natural human response (or AI response) to inventory and fears in the market.

a.      What will happen in 2024? How profitable will memory be in 2024?

b.      Why is NAND so much worse than DRAM?

c.      Why will the market fall again in 2026 (and yes, everyone will be surprised)

d.      How can we prevent this cycle IF we choose to (we haven’t chosen to yet, but we could)?

e.      How can you predict the next downturn and recovery?


3 The last FINFET  technologies are ramping now. How long will they last and how fast will GAA technologies ramp in 2025? Why will FINFETS dominate volume into 2028?


4 Intel is currently not a top 10 foundry today. They want to be the 2nd largest foundry in 5+ years.

a.      What do they need to do to break into the top 5 foundries?

b.      How are customers working with Intel today and when will we see success?

c.      How do we track Intel success or challenges?

d.      How will TSMC and Samsung respond 


5 Speaking of Intel…. Intel has to correct its cash flow and business model quickly

a.      How to pay for new fabs (what is ACTUALLY happening on NET capex vs gross Capex)

b.      Will Intel really have 5 nodes in 4 years …. Or is it really 3 nodes in 6 years? We tell you what will happen so you are not surprised when you see the news in 2024, 2025


6 Trade wars, Export bans, a foundry in every country. What will Fab world look like in 2030?


7 AI servers are the cool thing. But, by far, most servers being built today are classic standard servers. What does this mean for the datacenter market.  How to see if this changes in the next 3 years?

We have the info so that you can correctly predict the market as well. Contact us for an in person or zoom meeting. We will be at IEDM Next week and can set up a Zoom call

Mark Webb

www.mkwventures.com




Thursday, September 28, 2023

The Memory Recovery is Progressing. What happens next?

 In March and in May and in August we stated the cause of the memory crash and what will happen next. Everything is progressing as we predicted (check the presentations!)

-The Market bottomed out in what was the largest downturn in 20 years. 

-Bit shipments are now increasing

-Prices will start to increase in Q4

This is true for NAND and DRAM. NAND is worse off than DRAM in terms of oversupply and margins and the overall health of the market is still worse due to too many vendors being to aggressive on pricing and layer counts.

Micron announced results that still show major losses but those were exacerbated by 40% increase in NAND bit sales at discounted pricing. The underload will continue to weigh on Micron as well (see transcript of analyst meeting for more info and how they match our predictions.


What happens next?

1) Bit sales increase steadily

2) Prices increase 5% or so in Q4

3) Customers still have inventory, but knowing that the price will increase, plus the fear of missing out will lead to additional purchases. Plus we need to manage DDR4, 5, LP DDR, HBM, CXL.... getting the right product at the right time is tough (we will have shortages and surpluses at same time)

4) This will lead to perceived shortage and pricing will jump to profitability very quickly (Q1/2 2024). Perhaps we get a convenient power outage at a fab to help ! 

The questions that remain (we have predictions and ways for you to track this)

1) How much will the price go up? Will the upside pay for the downside?

2) Will the suppliers forget the last 18 months and oversupply yet again based on hype?

3) Will suppliers and customers make supply chain changes that allow them both to be successful


For details on the cause of the crash, the recovery, the changes needed, please visit out website or call/text us


Mark Webb

www.mkwventures.com










Thursday, August 17, 2023

Simplified HDD and SSD number for 2023

 With all the hype related to the downturn in compute and storage and the hype for future growth, I want to review what the breakdown looks like today and what it will look like in 2025


Couple introductions for our analysis

We focus on EBs shipped. Units are not useful other than to look at why EBs shifted. 

As the numbers will show, there are large differences between SSD and HDD. As a result, the key is to focus on high level view, not the details. We have data on details as well but it is more difficult to extract from company reports.

Enterprise/Cloud/Hyperscale ("datacenter" , non-client) is the market to focus on. HDDs are becoming niche market players in other markets. SSD have the best growth potential in datacenter. In this market servers typically have HDDs and SSDs attached.... but they exist at about a 9:1 ratio and there is no drive for widespread HDD replacement. 

HDD total EBs are shown. The data indicates that 85% of the HDD bits are Nearline (Datacenter).

eSSD is all datacenter SSDs (non-client)

Spoiler Alert: SSDs are growing. But they are nowhere near to becoming dominant overall. SSD EBs will not out-ship HDD EBs in the next 10 years. SSDs are dominant in the client space, but are a fraction of the bits shipped in the Datacenter space. 

In the period from 2019-2025. The INCREASE in HDD bits is higher than the TOTAL  bits shipped for ALL SSDs. HDD is not going away ... but is will be 90% nearline by then

2023 will be a downturn in total for all storage

2025 will be a peak and show the return to growth



Source: TrendFocus, Trendforce, Statista, MKW Ventures

Mark Webb

MKW Ventures Consulting

www.mkwventures.com






Monday, August 7, 2023

Mark Webb MKW Ventures Consulting at Flash Memory Summit

 


We will have multiple presentations at Flash Memory Summit Aug 8-10 2023

Text to set up meeting in person or by Zoom





First Presentation: BKMT-101-1, Tuesday 830AM Room GAMR1

2023 Memory Downturn, Recovery, Causes and Effects

We show the cause and effects of the current downturn in the memory markets. This downturn includes the first YoY decrease in bit sales in market history and massive losses by memory companies.

The data will show that this was not due to massive consumer market collapse but due to excessive inventory and unreasonable planning by memory companies. The end result will be that total bits shipped 2020-2023 will be the SAME amount we predicted in 2019. Unfortunately, the price collapse caused by excessive supply is lost income for the memory companies that cannot be recovered

When will it recover? What will the recovery look like? How can we take simple steps to prevent this in the future? We have the answers!

 --------------------------------------------------------------------------------------------------------------   

Second Presentation OMEM-201-1. Wednesday 830AM Ballroom E

Emerging Memory: The Calm After the Storm

“Emerging memories” are not really emerging. Some are here. Some have peaked and are now planning  their way out. Some are in research but are 10 years from revenue. They are also not “disruptive memories”. They are not replacing DRAM or NAND. Maybe NOR ... but no Christensen effect coming soon

We now have a calm after the “Optane”, “replace NAND”, “replace DRAM” storm. 

MRAM, ReRAM, PC-RAM, FeRAM are here today. They have applications where they are optimal and where they can grow. The technologies are being integrated into foundry processes. With Chiplet technology, you can put them on a module and flex the amount of memory used. What are the best uses and most cost effective applications? We present some ideas.

Revenue from “Emerging memory” will never be $30B+ as predicted by some other analysts. it will be 10x smaller than that. Specific applications will allow revenue to steadily grow to over a billion in the next several years.

 

Contact Mark to set up in person or zoom meetings to discuss Memory Markets and Emerging Memory

Mark Webb

www.mkwventures.com




 

Friday, May 19, 2023

Update on Memory Market Crash and Recovery (coming soon)

 


In previous blogs, we mentioned how and why the current crash happened and why





Normal Cyclical behavior that causes ups and downs based on pricing and perceived shortages. This happens every 4 years regardless of economy, wars, pandemics, ChatGPT, 5G. 

  • This was exacerbated by customers and suppliers thinking that DRAM and NAND bit growth were going to be higher than long term models of 15% and 25%
  • This was exacerbated by by PC market returning to normal TAM when PC OEMs had built inventory and assumed  300M+ unit TAM going forward
  • This was exacerbated by DC entering their digestion phase. 

End result was >12 weeks of EXCESS inventory. Typical downturn is 4-6 weeks excess inventory


This all led to the largest downturn in OVER 10 years. Perhaps the largest downturn ever

  • Bits dropped for DRAM and NAND YoY. These is essentially unprecedented
  • Prices dropped well below cost. 50% price drop in 9 months
  • All memory companies are losing billions in earnings and cash flow. 
  • Memory Capex has been cut by 50%


Memory will grow again, the market cycle will return up as it always does

The NEW issues will be that once we eliminate the perception of excessive inventory (real or not)

  • We need to get back to the long term bit growth curves
  • Capex has been cut and pushed out. wafer starts were cut
  • Pricing needs to increase 50% to break even and 100% to get us to "good margins" where we can afford growth. We have not been this far below cost in price on both NAND and DRAM for over  10 years

This means we will have a shortage overall, wrong part at wrong time (DDR4/5,  Mobile vs SSD NAND) and decisions made by suppliers on which markets to support and at what price

When?

  • The CQ1 numbers looked bad.... real bad 
  • CQ2 numbers look flat on bit growth/bad on ASP, this should be the bottom.
  • CQ3 should be flat with some customers starting to question whether then can get supply 
  • During CQ4 we will eliminate the perception of excess inventory (Perception matters). Then the correction to get to an acceptable price will begin. 

 

We have updates on how to track this on a weekly basis and what the new bit and price growth will be. We also have data on what will prevent this in the future. 

The next downturn will be in 2026 and everyone will be surprised. but it does not need to be as bad as this downturn


Updated ROUGH timeline



Mark Webb







Wednesday, April 26, 2023

When Will SSD Cost be Lower than HDD Cost? UPDATE

  This is the age old questions with the assumption that HDD will stop scaling and NAND will scale quickly

We updated all the numbers this week and the answer is STILL 2034! Maybe!

The HDD has a few tricks to pull out and NAND scaling will slow over time

The additional issue is that SSDs are now selling below Cost. So we added a note to show how price went from above cost to below cost in the last 12 months. HDDs are still cheaper than SSDs even today and the HDD is not yet selling below cost. SSD/NAND Pricing will improve in 2H 2023 and we will not need to factor in selling below cost for at least another couple years (hopefully)

The costs are for highest density and also do not include underload charges or other accounting related "one time items". We have all the details including one time charges and why NAND prices will improve .... contact us for more info.









Mark Webb

www.mkwventures.com




Tuesday, March 28, 2023

The Memory Market Crash and Coming Recovery



We have an Excel model predicts pricing and bit shipment effects and recovery. In our base case Excel model of memory cycles, we have simple assumptions and change from there based on actuals,






BASE CASE

1)     Suppliers keep 4 weeks inventory. Customers keep 4 weeks inventory.

3)     When there is a perceive shortage or above average bit growth, customers increase inventory or  double order to get to 8 weeks inventory

a.      They don’t tell suppliers, but the suppliers should know its happening

b.      Price is going up, so it make sense to buy ahead

c.      This accelerates the upturn and shortage

4)     Then the perceived shortage (real or fake doesn't matter) goes away

a.      Even if it goes back to normal, the customer needs to cut 4 weeks of inventory.

b.      Which creates more of a surplus, price drops

c.      Since price is going down and there is a surplus, a customer should cut inventory to 2 week (working inventory/JIT). Not doing this is financially wrong

d.      This leads to an excess inventory of 6 weeks.

e.      In some cases the supplier was building up inventory due to orders and cancellation of double orders. So there is 8+ weeks too much total inventory.

5)     Requirements from suppliers for next 6 months is now 30% less than it was during boom until it corrects.

a.      Prices are adjusted to incentivize people to buy bits they might not want. Price drops

6)     The graph of revenue and margins for last 15 years shows this very clearly

NOTE: in the above scenario, the economy is fine, Wars don’t matter, COVID doesn’t matter, it is simple perceived DEMAND and INVENTORY that caused the bullwhip effect. This is why memory is always cyclical.

TODAYS CRASH

This time was different: Due to supply concerns, and Taiwan concerns and hype about memory growth, Some MAJOR customers went from 4 weeks to 12 weeks inventory, instead of 8 weeks. When the shortage was relieved or didn’t materialize the result was shared with suppliers. That is why Micron and others very quickly cut production, which is a inefficient thing to do. They realized that this was far worse than typical and would take 1 year plus to solve if they kept bit growth moving. This is why it is taking longer to work off the inventory (should be 6months), why the price crash is so severe. Every memory company is losing money. It was caused simply by customer inventory.

When the smoke clears, I predict we will find that bit growth from 2021-2023 was DRAM 15%, NAND 25% just like everyone predicted beginning of 2021. No breakthrough apps, no black swan events, no impact of wars or invasions. Same as it ever was. The supply chain perceptions created a nightmare shortage and nightmare oversupply.

The issue is the next part of the story: how do we react??:

1)     Suppliers should go back to growing at long term rates.  15% DRAM, 25% NAND

2)     Suppliers should not add capacity or believe any demand increase, they are getting played by customers. AI will not change this. A slight permanent undersupply. Monitor end customers shipments

3)     When the oversupply ends, we often see 6-12 months of balance, then a future shortage. This time will be different, We will see <1qtr of normal this time. AND we need price to increase 20-30% just to make reasonable margins to cover development. Flattening of price will require 2 years to get back to profit.

4)     CAVEAT: this requires oligopoly where one company is not trying to gain market share.

 

SUMMARY

      + Oversupply is unprecedented in the last 10 years but will be corrected in ~Q3.  

      + Due to severe losses, and minimal inventory when oversupply is corrected, A shortage will appear quickly. Prices must go up to survive, supply growth needs to be constrained

      + Question 1: will the prices peak at 5% OM or 30% OM or something in between?

      + Question 2: will memory suppliers keep getting trapped into oversupplying or just grow at long term rates?

            We can monitor this for clients

Homework quiz: Which current memory end market is responsible for 90% of the Memory hype and <2% of the memory bit shipments. It is projected to be maybe <5% of bits by 2028. Hint: Ask ChatGPT whether DRAM Bit growth rate will suddenly increase in the next 5 years.





Mark Webb

www.mkwventures.com




 

Saturday, February 18, 2023

Intel ACTUAL Plans for Survival, Foundry, and Recovery (UPDATED)

 



Intel ACTUAL Plans for Survival, Foundry, and Recovery




Updated: Intel announced capex cut yesterday 2/21/23. This essentially will confirm my previous capex forecast made 1 year ago. 

  • Intel has made many aggressive commitments and strategy changes on new business model 

o   IDM2.0. Commitment to have large (#2 overall) wafer foundry business

o   Upgrade all existing sites for Intel 7,4,3. ADD 6 to 8 new fabs for foundry

o   5 nodes in 4 years. So far Intel have delivered 1 node in 2 years so now we need 4 nodes in 2 years (4,3,20,18)

·        Intel cash flow, market share loss and overall soft market is making those plans impossible

o   Intel historically does not lose money and does not have negative FCF. They do now

o   This makes funding fabs even with subsidies and JV partnerships difficult.

o   When the smoke clears we will see that semiconductor spending increased in 2022, and 2023 decrease for year as whole will be less than 5%. But Intel share losses and lack of growth in new areas has cut revenue and eliminated profit and positive cash flow

o   Intel is outsourcing more chips than ever to lower costs. Meteor Lake and Granite Rapids products are mostly non-Intel chips

o   End result is that Intel doesn’t need ANY new Fabs for Internal work and is not a priority for equipment vendors.

·        The public plan stated is unlikely to be what will actually happen. What we expect (Most of this was predicted in early 2022):

o   Intel will tool out and expand Arizona, Oregon, Ireland and Israel Fabs for Intel 7,4,3

o   Intel will complete and tool out 1 new Fab in Arizona by 2024. The other Arizona Fab will not be tooled out until demand is shown. This may lead to Intel payments to Brookfield Capital and tension in their relationship per contract details

o   Ohio will be “slow walked” with a new timeline and new milestones.

o   Other announced sites and factories will be “dependent on macro trends”

o   Macro has almost nothing to do with it. It will be dependent on whether Intel can develop a competitive GAA technology and have internal and foundry demand to fill the factory. Macro is a +/- 5% impact. Intel execution is +/- 30% impact on revenue.

o   Intel Foundry “Commitments” are <1 Fab right now. It is not clear how they increase commitments

·        We show what potential scenarios are with the most likely scenario detailed for you

o   Intels new Capex plans aligns with the fab roadmap strategy we will show

·         While Intel IDM2.0 strategy is not likely to be successful based on current data, Intel will recover to be a successful company again by focusing on where they ACTUALLY add value (This will be just like the IBM Journey 10-20 years ago)

·        Most importantly, we tell you how you can track progress and changes to strategy yourself over time


JOIN OUR ZOOM MEETING. We discuss details and questions on what will happen in the next 1-5 years

Intel Future: Tuesday 2/21 3PM Pacific Time  (30 Minutes+questions)

   828 0396 9706 Passcode 021289

Intel Future: Wed 2/22 5:30PM Pacific Time (30 Minutes)

817 1243 6496 Passcode 773007

 

We are free to meet by zoom or in person in San Francisco. text us for a appointment

 

Mark Webb

MKW Ventures Consulting

www.mkwventures.com




Thursday, February 16, 2023

Memory Market Collapse: Cause, Timeline to recovery, What the new margins will be

 


Mark Webb, MKW Ventures Consulting will be at ISSCC Feb 19-23

We will be having meetings with people and providing updates to partners and clients. 


We are also hosting two different Zoom meetings (with multiple times) on two hot topics in the industry right now. see times below


 Memory Market Collapse. The worst downturn in 10+ years, but it will recover

Topics:

Memory is always up and down. We can show how cyclical behavior is actually predictable and can be modeled with an excel macro. What is different this time is the magnitude of the drop. It has become one of the worst downturns of all time. 

ASPs down 30-40%. Bits are down YoY (this is extremely rare)

Revenue drop combined with previously committed Capex causing huge free cash flow problems

When the smoke clears we will find that a 10% end user consumer drop from forecast caused a 50% revenue drop. Why did this happen and how could we have prevented it.

When it will recover (It will recover) .... and when the next shortage will be (it will be a major shortage)

Reminder: Memory and other semiconductors markets are COMPLETELY different in behavior. We can discuss how to treat those markets differently

When will we recover? Why was this so bad? What will the new margins be after recovery? Bring questions, we have answers


Zoom:

Memory Market: Tuesday 2/21 2PM Pacific time (30 Minutes plus Questions)

   828 0396 9706 Passcode 021289

Memory Market: Wed 2/22 5PM Pacific Time (30 Minutes)

817 1243 6496 Passcode 773007


We also have a second Zoom call (follows the time of the Memory call). It will cover

Intel ACTUAL Plans for Survival, Foundry, and Recovery



We are free to meet by zoom or in person in San Francisco. text us for a appointment


Mark Webb

MKW Ventures Consulting

www.mkwventures.com




Friday, February 10, 2023

Mark Webb MKW Ventures at ISSCC Feb 19-23


 Mark Webb, MKW Ventures Consulting will be at ISSCC Feb 19-23


We will be having meetings with people and providing updates to partners and clients. 




To schedule a meeting please text us at the phone number below

We are also hosting two different Zoom meetings (with multiple times) on two hot topics in the industry right now. see times below


1) Memory Market Collapse. The worst downturn in 10+ years, but it will recover

Why it was totally preventable

Why does a 10% consumer demand drop lead to a 50% revenue drop

When it will recover (It will recover) .... and when the next shortage will be (it will be a big shortage)

Reminder: Memory and other semiconductors markets are COMPLETELY different in behavior


Memory Market: Tuesday 2/21 2PM Pacific time (30 Minutes plus Questions)

   828 0396 9706 Passcode 021289


Memory Market: Wed 2/22 5PM Pacific Time (30 Minutes)

817 1243 6496 Passcode 773007



2) Intel ACTUAL Plans for Survival, Foundry, and Recovery

Intel has made many aggressive commitments, plans, and strategic changes

Intel cash flow, market share loss and overall soft market is making those plans impossible

The public plan stated is unlikely to be what will actually happen.

We show what potential scenarios are with the most likely scenario detailed for you

Most importantly, we tell you how you can track progress and changes to strategy


Intel Future: Tuesday 2/21 3PM Pacific Time  (30 Minutes+questions)

   828 0396 9706 Passcode 021289


Intel Future: Wed 2/22 5:30PM Pacific Time (30 Minutes)

817 1243 6496 Passcode 773007


We are free to meet by zoom or in person in San Francisco. text us for a appointment


Mark Webb

MKW Ventures Consulting

www.mkwventures.com