• The “ouster” of Pat Gelsinger means that Intel will almost certainly be split up and could be absorbed into other companies meaning that the once all-powerful Intel brand may soon be just a memory.
• It is also a further sign that Intel lacks the corporate culture that would enable it to fight for its position after decades of cruising along at the top without having to do very much other than get the manufacturing right.
• This is also a gift for Qualcomm, Nvidia, AMD and MediaTek to come and take its business while Intel is tying itself in knots with its internal issues.
• Pat Gelsinger is “retiring” from Intel which I think in plain English means that the board forced him out after it lost confidence in his ability and his plan to return Intel to its former glory.
• This has happened very suddenly and with no successor in the wings, it is clear that this was not planned, supporting my view that Gelsinger has been forced out.
• This means that the plan to invest heavily and return Intel to the front line of semiconductor manufacturing with foundry and design remaining under a single roof is now out of favour with the board.
• This means that Gelsinger’s successor will most likely press ahead with a plan that breaks Intel up into separate pieces which I think spells the end of Intel as we know it.
• Intel’s greatness was based on the combination of manufacturing excellence and its ownership of the x86 design that had dominated PC and data centre processors for decades.
• With both of these under attack from all sides, its only real hope to survive was to make a success of Pat Gelsinger’s plan and while it has proved to be much harder than expected the migration to leading edge is on track.
• The problem has been the foundry business which was split out as a separate business earlier this year leading to the market realising just how much money the division is losing.
• Foundry is not expected to be properly competitive until 2026 when 18A should be running in volume but the board has clearly decided that the Gelsinger has run out of time.
• The real problem at the heart of the problems as I see it is cultural, as demonstrated by its response to the X Elite from Qualcomm.
• Looking at the recent trajectory of Intel’s chips prior to the X Elite, one can see incremental improvements keeping in step with AMD but not closing any of the gap with Apple’s M series.
• Then suddenly when its house is suddenly under threat, it awakes from its slumber and produces a chip (Lunar Lake) that offers a significant performance improvement and is far closer to the X Elite or M series than one would have predicted looking at its history.
• This suggests that Intel’s culture is reactive and somnolent and that any get and up and go that the company once had has long since gone after decades of cruising along at the top.
• Its ability to produce Lunar Lake demonstrates that if it had taken the initiative, it could have prevented AMD’s resurgence as well as made life much harder for Qualcomm and the Arm crowd in the PC market.
• A company with this attitude could have made a success of Gelsinger’s plan and his failure is his inability to reignite Intel’s competitive spirit.
• This is a truly dire state of affairs because fixing this malaise will be very difficult but not impossible as Satya Nadella has proven (although things were not nearly as bad at Microsoft when he took over).
• With Intel’s future very uncertain and its once great engineering culture in poor shape, Intel is easy pickings for its resurgent competition.
• I expect to see Qualcomm, AMD and MediaTek redouble their efforts in PCs and the data centre as Intel’s entire business appears to be up for grabs.
• I have often said that at 10x earnings, one should shut one’s eyes and buy Intel but because of the falling EPS estimate the valuation has never gotten there.
• With the company in this much trouble, there is no price at which I want to buy it.

• The vultures are hovering around Intel and while Intel will make a tasty and nutritious meal for some, I think Intel would end up sticking in Qualcomm’s crop and give it a bad case of indigestion.
• Multiple sources are now reporting that Qualcomm has had discussions with Intel about a possible acquisition which leads me to think that the discussions have happened, but nothing more.
• Furthermore, there are other possibilities for Intel as Apollo is proposing an investment into the company that would provide badly needed cash and improve confidence that Pat Gelsinger is on the right track.
• At a very high level, it is possible to see how Qualcomm and Intel could fit together as Qualcomm designs chips and Intel has factories.
• The problem is that Intel has many other things that don’t fit so well and having known both companies for many years, their corporate cultures are also very different.
• I see the problem areas as:
  • First, x86: which is under assault on all fronts from Arm and accelerated computing.
  • Qualcomm has spent 20 years competing against x86 both in mobile and now far more successfully in PCs where it is on track to take a lot of market share.
  • Consequently, it makes no sense to become the owner of a fading titan when one is in a good position to partially or completely displace it from the market.
  • I see no reason to dilute shareholders to purchase revenues that could very well be coming one’s way anyway.
  • Second, Fabs: where Qualcomm has already established excellent relationships with TSMC, Samsung and GlobalFoundries and knows exactly how to design its chips to be optimally manufactured on these processes.
  • Intel is behind on the leading edge (although it is catching up) and redesigning its chips to be made by Intel (and only Intel in an acquisition) sounds like a risky proposition to me.
  • Even if Intel recovers and becomes the leading semiconductor manufacturing company again, Qualcomm will still be able to access its fabs through Intel Foundry.
  • Taking on this kind of fixed asset ownership has never been part of Qualcomm’s DNA, and owning fabs like these will expose Qualcomm to a level of cyclicality that it has never had to deal with before.
  • Third, Dilution: where this will be an expensive acquisition.
  • Qualcomm would almost certainly pay for Intel with shares which would result in dilution for existing holders of Qualcomm’s stock (of which I am one).
  • Intel’s share price has cratered but so have its fortunes and this means that the shares are not cheap even at $22 per share.
• The one part of Intel that Qualcomm could really use is its relationship with the distribution channel.
• In the PC ecosystem there are vast numbers of participants and ensuring that your products work perfectly with everyone else’s and ensuring that everyone else wants to make their products work with yours is a massive challenge.
• Intel has been at this for over 30 years and so its knowledge, experience and relationships in the ecosystem and the channel are second to none.
• This is how it launched a new chip with 80 designs in the pipeline while Qualcomm despite Microsoft’s wholehearted support could only manage 20.
• Qualcomm has been at this for a year or so and so it has a long way to travel before it can rival Intel in this area.
• This is the one part of Intel that I think could be of great use to Qualcomm but crucially, there is no need to acquire the company to get it.
• I have noticed that a number of the critical personnel have recently left the company, and I suspect that this was not due to the current 15,000 headcount reduction that is ongoing.
• I will not be surprised in any way if they pop up at Qualcomm and are tasked with the recreation of the channel that Intel has developed over many years.
• Hence, I see no reason for Qualcomm to acquire Intel as it looks like it will be able to get the bits it needs without having to also take on the baggage.
• I have always said that at 10x PER, Intel represents good value, but the fundamentals have unravelled faster than the share price which is why I have never been tempted.
• Qualcomm on the other hand has none of this baggage, is opening up new markets for its chips and has opportunities to gain share in markets that are not included in consensus estimates.
• This is why I own Qualcomm and not Intel and will be very happy for it to remain that way.

• Intel has formally launched its answer to the Qualcomm X Elite and AMD’s Ryzen AI 300 series and if its benchmarks are to be believed, the fight for the laptop market will be closer than anyone expected (including me).
• Lunar Lake was originally supposed to be released in 2025 but with the extreme pressure exerted by Qualcomm (and now, AMD), this was rolled forward into 2024 with devices expected in Q4 2024.
• On 3^rd September, Intel released the benchmarks for Lunar Lake Core Ultra 200V series and announced that the laptops are now available for pre-order from its usual partners and will begin shipping to customers on September 24^th.
• The benchmarks are impressive given Intel’s recent performance but will need to be verified when devices are in the hands of 3^rd parties.
• I suspect that reviewers are already testing these products but as usual, they are likely to have been embargoed until September 24^th.
• Furthermore, Intel claims to have used production rather than prototype devices for its testing implying that these figures should be a reasonable representation of reality.
• Intel is claiming “historic x86 power efficiency” and that the Core Ultra 200V series achieves 1.2x improvement in performance per watt than the Qualcomm X ELITE x1e-80-100 which is 2.3x better than Intel’s previous generation.
• This translates into 20.1 hours of battery life using a well-known Office productivity benchmark, 10.7 hours on a team call which beats Qualcomm’s 18.4 hours on Office productivity but loses out against Qualcomm’s 12.7 hours on a Teams call.
• Intel also spent time calling out the 2.2x improvement in power efficiency on gaming versus the previous generation, highlighting the 23 games that it tested where the Qualcomm chip did not run at all and claimed a 16% improvement over AMD’s AI 300 series.
• On the games that would run on Qualcomm, it claimed a 68% advantage and went on to hammer the competition on its AI performance.
• Here, it is claiming a maximum of 120 TOPS when combining resources from the CPU, GPU and NPU and highlighted that it retains support for FP16 instructions while AMD and Qualcomm top out at INT8 for AI inference.
• However, of Copilot+ there was no mention because Microsoft’s x86 version of Copilot+ is still not ready and is not expected to be released before November 2024.
• This clearly demonstrates where Microsoft’s priority for this generation lies, and so I think that Qualcomm is going to have plenty of support when it comes to the long and difficult job of fixing all of the things that so far do not work with Windows on Arm.
• Assuming that Intel’s benchmarks prove to be reasonably accurate (and I have no real reason to think that they won’t be), then the stage is set for a large increase in competition in laptop processors which is excellent news for the consumer and Microsoft.
• Since Apple launched its groundbreaking M-series of laptop and desktop processors, there has been a slow but steady influx of higher-end users to Apple, which has forced the Windows community to react.
• Qualcomm’s entry forced both AMD and Intel to react and the result is a large jump in performance and battery life which is only going to improve further as competition intensifies.
• The big winners are Microsoft and the consumer but even if the benchmarks are correct meaning that the Arm advantage is much less than expected, Qualcomm still has an ace up its sleeve.
• That ace is price because, on a straight comparison, the Qualcomm-powered devices are significantly cheaper than either AMD or Intel.
• This means that even if one does not care about AI in Copilot+, Qualcomm-powered devices offer more bang for the buck.
• The third generation of Windows on Arm is a night-and-day improvement over the last time this was tried, and the commitment is there to take this to the mainstream.
• The vast majority of the stuff that most users care about works flawlessly and a lot of the long tail of apps that use the emulator work well enough so that the user does not notice the emulation.
• Generally, anything that is written directly to hardware (like games or VPN apps) that have not been ported struggle on Windows on Arm but this time, the will appears to be there to fix these issues.
• Hence, Qualcomm is unlikely to take any share in gaming laptops for a while but this was never the intent at the moment and it is important to remember that this is just the first generation.
• I think the stage is set for Qualcomm to take share in laptops, especially in the mid-to-high-end market, where usage is less specialised and price is a factor.
• This should provide it with a platform of volume from which it can address the areas where Windows on Arm still needs attention to the point where everything works equally well on x86 and Arm.
• This will take some time but almost none of this is included in the current forecasts for Qualcomm.
• Hence, I see upside in revenues and profits in Qualcomm compared to current expectations meaning that it remains one of the most inexpensive ways to play the AI boom.
• I have a position in Qualcomm and remain very happy to hang onto it regardless of the current market volatility.

• For a first generation, things are looking very good, but it is clear that when one gets into the weeds there are a few loose ends that will need tidying up to really put an end to x86 in computing.
• The Copilot+ PCs are in the hands of the reviewers and by and large, the reviews are good but there are problems here and there that might make users wait for generation 2 before taking the plunge.
  • First performance: which is generally excellent when the app or test in question is running natively on Arm.
  • Tomsguide’s tests (see here) demonstrated that most of the claims made for performance are broadly accurate.
  • The X Elite (12-core version) comfortably beats Intel in multi-core tests and in some cases, beats the M3 and others match it.
  • The differences from one manufacturer to another depend on how the chip has been tweaked in terms of the firmware how the PC has generally been built and what it has been optimised for which is why there are differences with the same silicon.
  • It is important to note that the X Elite everyone is testing is not the fastest one that will be offered in this generation and so one would expect to see a moderate boost to these figures when that chip is available in devices.
  • On regular usage tests with the usual day-to-day apps and productivity, the feedback is universally positive but there is more mixed feedback on video editing.
  • For example, Adobe’s apps are working flawlessly but testers have reported some issues with Davinci Resolve and Blender.
  • Second, emulation which is for apps that have yet to be ported from x86 to Arm where they can run natively.
  • Once again, the vast majority of the time, the emulated apps work just fine as long as they are not beta versions but there have been issues reported with gaming.
  • For example, on the Windows Laptop, the Xbox app only supports cloud-based gaming while in the Windows Store, many of the games are not supported.
  • Games are downloadable from the usual routes like Steam, but many of them appear not to run well or at all.
  • One will never see this long tail 90% of the time, but Intel and AMD are likely to jump on with alacrity.
  • The obvious answer is that these are not machines designed to run games, but I am certain that these issues are being addressed and will improve a lot in the coming months.
  • Third, battery life which is one of the big selling points of these devices.
  • The claim on these devices is that they can run for 20 hours + in between charges and while this is demonstrably true, these tests are not often very good models for real-world usage.
  • In most tests for general usage, one can get a full day of usage out of the Arm-based PCs without any real difficulty, giving them a solid advantage over the best of what x86 currently has to offer.
  • However, when comparing to Apple silicon some areas for improvement did emerge such as battery drainage on standby and power consumption when the processor is running flat out.
  • The good news here is that all these issues are largely software-based, and so future tweaks to firmware should be able to rectify a lot of these shortcomings.
• The net result is that Arm on Windows is broadly living up to its promises in the areas that really matter, but Microsoft needs to redouble its efforts to ensure that the loose ends are tidied up if it wants to halt the bleeding to the Mac.
• It also highly relevant the X Elite is a 1^st generation product which almost always has issues like these and it is being compared to Apple’s silicon which is currently in its 4^th generation and has had much more time to iron out the kinks in the product.
• Hence, I think that these devices will appeal quite strongly to anyone who does not care very much about gaming (which is who Copilot+ is targeted at).
• However, there are clearly some loose ends that need to be tied up and it looks very much like these can be dealt with via a software update as opposed to having to buy a new machine.
• There is still some work to do before Arm can knock x86 out, but the foundation from which it can be done has been established.
• In short, a winning proposition but it needs some spit and polish to really start firing on all cylinders.

• The key theme of Computex was the arrival of generative AI on the PC and while most attention was on Copilot+ running on the NPU, Nvidia was quick to remind everyone that if you want real AI horsepower, you need to use an Nvidia graphics card.
• Copilot+ is a runtime that uses a number of smaller language models (SLMs) to provide specific generative AI services on the PC that developers can write to and incorporate into their apps.
• Typical examples are photograph enhancement, image generation from a sketch, real-time translation and so on.
• The Copilot+ runtime runs entirely on the PC and is designed to run on a specific processor called a neural processing unit (NPU) which is designed to run inference on neural networks both quickly and efficiently.
• Microsoft’s specification for this is an NPU with at least 40 TOPS which Qualcomm is already shipping with Intel and AMD hot on its heels.
• The NPU is demonstrably the most efficient way to run Copilot+ and there is a demonstration by Microsoft that shows this.
• Microsoft shows two PCs running Copilot+ one on a Qualcomm NPU and the other on a Nvidia 4060 chip with the Nvidia PC measured at 50+°C while the Qualcomm X Elite device is measured at 30°C (see here) indicating that the Nvidia device is consuming far more power meaning that the battery life will quickly be drained.
• This indicates that for SLMs, the NPU is the way to go, but as always, there is more to this story than meets the eye.
• Nvidia’s 4060 is capable of running nearly 8x (353 TOPS) faster than the NPU in the X Elite and so it is not a surprise that it consumes 160 watts of power under load explaining why it gets so hot in the task above.
• Consequently, Nvidia was demonstrating a whole series of AI tasks at Computex that require far more power than 45 TOPS almost all of which were running on desktop PCs as opposed to laptops.
• It is becoming clear that the kind of AI tasks one is aiming for will determine where they run with larger models running on Nvidia and smaller models on the NPU.
• This is very similar to the rivalry between x86 and Arm that existed in computing devices for many years where top performance required x86 which consumed a lot of power, but efficiency and battery life were far better on Arm albeit with lower performance.
• This situation was put to bed several years ago by Apple which demonstrated that Arm could do both which has initiated the greatest threat to the x86 processor in its history.
• The trend at the moment (rightly or wrongly) is for AI models to get bigger as it is believed that this is the best way to make them perform better.
• Followed to its logical conclusion, this implies that as AI becomes more integrated with general computing tasks, the performance of the processor executing the task will need to keep rising.
• Furthermore, the fact that the really big tasks are likely to be carried out on desktops that have no batteries means that power consumption will be much less of an issue.
• Hence, it looks like tasks on the NPU will dominate in devices where battery life is important and Nvidia will have a strong position in high-performance tasks where power is less of an issue.
• I would expect the NPU to erode Nvidia’s position from the bottom over time but at the same time, if the models continue to get bigger, there will be plenty of space in the desktop market for Nvidia to sell into.
• I suspect that the moment when the NPU can offer Nvidia performance at a fraction of the power consumption is far away and so this status quo is likely to exist for some years to come.
• This is now a rounding error in Nvidia’s financial performance as 90% of revenues are coming from the data centre and so I do not expect this issue to have any impact on Nvidia’s fundamental performance.
• I also do not see Nvidia impacting the case for running Copilot+ on the NPU or the case for Windows running on Arm processors in the laptop market.
• Qualcomm is going to have this market to itself for a few months as Copilot+ for x86 is not yet ready to be released and so the Intel and AMD laptops will not offer Copilot+ out of the box but it will be available as a future upgrade.
• This is what “Copilot+ ready” means.
• What impact this has remains to be seen but I still think that Windows on Arm has the possibility to take a large amount of share from x86 and the best way to play that remains Qualcomm.
• I own Qualcomm and am quite happy to keep sitting on it.

• Mediatek rolled out the big guns at its keynote with both Arm and Nvidia CEOs joining Rick Tsai on stage and although it did not say very much, it certainly made its presence felt.
• The main theme of the session was that MediaTek intends to broaden its portfolio to include AI acceleration in the cloud as well as hinting that it would follow Qualcomm into PCs and AI at the edge.
• This is where its partnership with Nvidia which started in automotive becomes very relevant.
• This partnership exists as there is a very good fit for these two companies in automotive semiconductors.
• This is because the digital cockpit and ADAS domains are merging and one needs both CPUs to run the cockpit and GPUs to run machine vision for ADAS.
• Nvidia retired its Tegra processor quite some time ago and so there is a very good fit between Mediatek and Nvidia to jointly address this space.
• However, it does not stop there and Mediatek also intends to address the custom niche for AI accelerators in the cloud which is where companies want a customised system to train and run their AI in the cloud.
• Counterpoint Research has forecasted that this will be worth $45bn in 2028 and it looks to me like Mediatek will address this using Nvidia IP.
• I suspect that Nvidia will take most of the value from this arrangement, but given Nvidia’s strength in this space, there is going to be plenty left over for MediaTek.
• Having both Jensen Huang and Rene Haas on stage at its keynote was a big PR coup and its association with Nvidia is going to ensure that it will be able to open the door to any potential client it wants to address.

Floor vs keynotes: boxes out of the door.

• While the keynotes were all about how the world of computing will be changed by AI, the mood on the floor of the exhibition was quite different.
• Here, there are two halls and 5 floors full of companies that offer components, subsystems and accessories that are used to make computers of all shapes and sizes.
• The stands of all of the major players have a section which is related to AI PCs, but the vast majority of the show remains about developing and selling bits and pieces that enhance PC performance or customises them for a specific use case.
• There are also a large number of players that offer components that make desktops look attractive which I suspect sell in very low volumes.
• This contrast clearly demonstrates that the AI PC is a brand-new concept and one that is largely untested on the PC buyer.
• The demonstrations are excellent meaning that if the software lives up to its billing, then there is scope for an uplift in PC shipments as well as a halt of the bleeding from Windows to Macintosh.
• This is likely to be decided in H2 2024 as the first AI PC devices are shipping on June 18.

• Intel came out of its corner swinging taking on both Nvidia and Qualcomm in a keynote that was really all about how its latest processors stack up against the new upstarts.
• Intel was quick to refute Nvidia’s claim that Moore’s law is dead and if any of its data is accurate, there is some substance to this claim.
• This was on display with its latest data centre CPU Xeon 6 where eBay and SAP are supporting Intel’s claims.
• eBay claims that the Xeon gives a 25% improvement in performance per watt while SAP was able to squeeze out a 60% saving over the previous generation of Xeon.
• Intel also took a shot at the Nvidia H100 claiming that Gaudi 3 is 40% faster than Nvidia’s top-selling H100 processor and can deliver 2.3x the performance per dollar spent.
• The problem here is that the Hopper is no longer the benchmark as Nvidia launched Blackwell a couple of months ago against which Guadi 3 will be much less impressive.
• I suspect that Gaudi 3 will still be cheaper when it comes to performance per dollar when compared to Blackwell but this is also because everyone is so desperate to get their hands on Nvidia’s products that it can earn nearly 80% gross margins which is almost unheard of in semiconductors.
• Gaudi is a good option for the few that don’t want to use Nvidia or want to dual source, but it doesn’t have CUDA and as such, I don’t see it as a massive competitive threat.
• In AI PCs, Intel also came out swinging at Qualcomm which will be pretty gratifying for Qualcomm that Intel considers it to be such a threat.
• On the last two occasions that Qualcomm has taken a shot at this market, Intel didn’t even bother to acknowledge that the threat existed.
• It certainly seems to be the case that it is third time lucky for Windows on Arm.
• Here Intel claimed that the Lunar Lake CPU, GPU and NPU can all outperform the X Elite, which is credible, but crucially, Intel neglected to mention at what level of power consumption.
• Unlike its peers, Intel was brave enough to make power consumption comparisons against Meteor Lake, it said that Lunar Lake could achieve a 40% SoC power saving.
• Qualcomm’s test claims a 65% saving against Meteor Lake and so it’s a pretty safe bet that the X Elite will fare better on performance per watt.
• However, the question is how much and here there is no way to know until the devices get in the hands of consumers and testers and we see them work in real-world situations.
• Purely on the numbers if Meteor Lake is consuming 100 units of power, then Lunar Lake is claiming 60 and X Elite 35 implying that X Elite will still offer substantially longer battery life than Lunar Lake.
• It is clear that Intel is not going down without a fight, but Qualcomm is coming to market with Copilot+ PCs first and as such it is winning the PR battle against the x86 camp.
• Against Nvidia’s public presence juggernaut, Intel also faces an uphill battle meaning that this Computex has found other champions.
• Intel still has everything to do.

Qualcomm: Swinging for the fence.

• With branded shoes and a flashy t-shirt, Copilot+’s most enthusiastic fan made his Computex debut bringing all the PC makers who matter on stage to extol the virtues of their new AI PCs as well as Qualcomm’s new prowess as a PC chipmaker.
• Most of the announcements were made at the Surface event a couple of weeks ago but this was the first time we got to see what Qualcomm actually has to say.
  • First, performance and battery life: which were front and centre of every comment which is really refreshing to see.
  • From my perspective, the real story with the X series chipset is the ability to match its competition on performance but then offer double the battery life.
  • Asustek’s (see below) press conference brought this home as most of its PCs are on AMD’s AI processor (see below) where battery life was not mentioned once.
  • This makes me think that AMD (and maybe Intel) are going to be in real trouble on this metric and Qualcomm is greatly helping its proposition by hammering this message again and again.
  • This is all over its marketing campaigns including the amusing advertisement of a Mac user ordering a Copilot+ PC that was shown as the keynote closed.
  • Second, AI PCs where Copilot+ is a series of small language models (SLMs) that run in the background and allow the Windows experience to be far more intuitive than it has been before.
  • App developers can use these SLMs in their apps to enhance their functionality and there was plenty of this demonstrated both on video and in the booths.
  • This is being hailed as a rebirth of the PC and to be fair to its proponents, if it works as well as it is being demonstrated, then this should be a driver of sales.
  • The Copilot+ proposition has yet to be properly tested in the real world and it is impossible to tell just how good this is going to be and how much consumers are going to demand it.
• Crucially, Qualcomm does not need the AI PC to be a smash hit to gain share as it is leading with performance and battery life which are enough to gain a lot of share on their own in the medium term.
• The only blot on the landscape at the moment is the ongoing litigation between Arm and Qualcomm which is preventing them from co-marketing this proposition.
• This is unfortunate as I think that together, their joint message would have more impact and benefit both companies.
• Regardless of this issue, Intel and AMD still have a lot to do.

Arm: 50% in 5 years

• Arm put on a confident display and just like everyone else, it is moving to lock down the AI ecosystem but a great opportunity to maximise the impact of Windows on Arm is being missed as a result of the ongoing litigation between Arm and Qualcomm.
• Much like Jensen, Arm spelled out its history and went on to lay out its proposition to further increase its presence in computing.
• Here, Arm also has a proposition on top of power efficiency for developers that it refers to as Arm KleidiAI.
• KleidiAI is an API that plugs into all of the most popular AI frameworks that developers use so that they don’t have to worry about what configuration of Arm cores or processors are being used underneath.
• This will make it much easier for developers to create AI services for all of the device categories that use Arm further increasing its appeal.
• Alongside a new 3nm processor for laptops, smartphones and other devices, Arm stated in an interview that Arm could gain 50% market share in Windows within 5 years.
• This is precisely the kind of statement that should have been made on stage with Qualcomm because if this is even remotely accurate, there is a meaningful upside to both Qualcomm’s and Arm’s medium-term estimates.
• This leads us again to the ongoing litigation where I think that it is in both companies, interest to settle and start working together in the open once again.
• This plays directly in MediaTek’s favour as I suspect that MediaTek will address the laptop market with Arm designs where it is likely to fare particularly well in China.
• Arm now has pretty good support for its valuation at around $100 meaning that as its estimates rise over the next couple of years, the share price will have space to go up.
• I think there are alternatives that offer better financial value for investors but Arm’s trajectory looks fairly secure for the short to medium term.

AMD / Asus: Not a peep about battery life.

• AMD joined Intel in launching a competitor to the X Elite but when it came to launching devices, Asus was strangely silent when it came to mentioning just how much battery life these devices would have.
• AMD’s new chip is the Ryzen AI 300 series of which the Ryzen AI 9 HX370 will be used in a number of Asus’s new laptops.
• This has 12 CPU cores, 16 GPU cores and a 50 TOPS NPU which matches the specification required to run Copilot+.
• Copilot+ is a run time that includes a number of small language models (SLMs) each of which is capable of performing a certain function like graphics enhancement, generation or the Recall function.
• This means that these SLMs need to run in the background to provide a good user experience raising the question of just how much power they will consume.
• AMD, Intel and the OEMs have yet to offer a straight answer to this question and with Copilot+ for x86 not yet available from Microsoft, Qualcomm is likely to have this market to itself for a few months.
• The real test will come when laptops are available from all three companies and real tests can be run to compare how much power each of them consumes when running Copilot+ in the background.
• Given the unwillingness of AMD and Asus to discuss this issue, I suspect that Arm is going to come out ahead and assuming that the software works as well as promised, the stage is set for the status quo in PC processors to be disrupted.

• Apart from AI, Microsoft Build is likely to see the main launch of Qualcomm, Arm & Microsoft’s 3^rd attempt at addressing the Windows laptop market, and this time around, it has by far its best chance of success.
• This will then be followed by the PC trade show of the year, Computex (June) which is likely to see Qualcomm go all out with the rest of the PC ecosystem to demonstrate that Windows running Arm is a viable alternative to x86.
• First out of the gate is likely to be Microsoft which is holding a Surface event on May 20^th, the day before Build kicks off.
• A 13” and 15” Surface Laptop are expected to be launched as well as an Arm-based version of the Surface Pro 10 which launched with Intel in March this year.
• All of these are expected to be based on Qualcomm’s X Elite processor which in raw performance and power consumption, sits just behind the Apple M3 processor.
• I also would not be surprised to see some other PC OEMs launch at Microsoft Build, but I suspect that the biggest push will come at Computex in a couple of weeks.
• Asus is also expected to launch a device on Monday and Dell is expected to launch three variants of its XPS 13 in June all of which will feature a Windows-on-Arm variant.
• I suspect that all of the Windows laptop OEMs will have a shot at this market, but this is by no means a guarantee of success even if the promise of the same performance as x86 but double the battery life is fulfilled.
• There is a caveat that has sunk all of the previous attempts to build a Windows-on-Arm laptop in the past.
• All of the previous attempts have come with limitations which have meant that the user experience compared to x86 variants has been poor meaning that hardly anyone has wanted to use them.
• This is all about the software and the problem here is when users download anything for Windows, they expect it to just work.
• Apple managed the transition by having an emulator (Rosetta) that could run x86-based apps on its Windows device with no noticeable impact on the user experience or the battery life of the device.
• This is the bar that Microsoft, Arm and Qualcomm have to get over and their previous failures have left the commentariat highly sceptical about this 3^rd attempt and rightly so.
• However, the evidence has been building for some time that what Apple did in Macs is indeed possible on Windows and that it has its best chance of success this time around.
• Qualcomm, Arm and Microsoft have been acutely aware of the shortcomings of the past and a lot of effort has been put into getting the emulator and the ecosystem right.
• Qualcomm has gone so far as to claim that its emulator was offering better performance than Rosetta and at lower power, which if this plays out in the wild, gives it an excellent chance of success.
• This is because from a user’s perspective, anything he or she downloads or installs will just work and he or she won’t be able to tell when the emulator is running compared to when it is not.
• The other issue is the long tail of component makers which needs to be good at launch but will become much more important when the X Elite moves into desktops.
• There is an almost infinite number of permutations and combinations of hardware that can go into a desktop (and to a lesser degree laptop) all of whom have more than 20 years of experience writing drivers for Intel.
• All of this now needs to be cross-tested and prepared for Arm and Qualcomm has been making a big effort on this front also.
• Apple’s vertical integration meant that it never had this problem which is why Qualcomm, Arm and Microsoft have had to go above and beyond what Apple did when it made the transition.
• The fruits of this effort should be on display at Computex which is going to provide the first real view of whether Windows-on-Arm is a viable proposition.
• If this does work, then there is significant financial upside for both Qualcomm and Arm who will be taking market share from Intel and AMD in a 250m unit market with very high processor prices.
• This time around, I think that the chances of success are good as there is far more commitment to make this work than I have ever seen before.
• Hence, I think it is more likely to work than not, but I remain to be 100% convinced which is why I am visiting Computex in person for the first time.
• This is also a major reason why I am holding onto my shares in Qualcomm as this, combined with automotive, could easily drive further estimate upgrades from the market which will keep pushing the shares up.
• This will also undoubtedly help Arm, but because the multiples on the shares are so much higher, the absolute level of upside on Arm compared to Qualcomm is probably lower.
• I have Qualcomm and remain happy to sit on it.

• While the technology world continues to jump up and down over generative AI, the most consequential challenge to Intel and AMD’s dominance in laptop processors is almost ready for launch and this time it might just work.
• Intel has more than enough reason to scoff at the prospect of losing market share in laptop processors to Arm-based rivals as on both of the two previous occasions this has been tried, it has been a miserable failure.
• For decades, the x86 processor has dominated the laptop market as it was thought that when it came to choosing a processor architecture (x86, Arm, RISC-V etc) one had to choose either performance or battery life.
• However, this all changed in 2020 when Apple launched the M1 processor that conclusively proved that it was possible to have professional-grade performance on an Arm-based processor with excellent battery life.
• Furthermore, it quickly became clear that Apple would not have a lock on this ability as it emerged that it was suing Nuvia, a start-up aimed at creating high-performance Arm-based processors.
• This was a sure sign that the changes that it had made in its processors were not patentable, which is why it sought to stop competition using the legal system.
• However, the lawsuit never went anywhere, and Qualcomm ended up buying Nuvia so that it could replace the Arm-processors that it had been using with its own designs.
• I have long thought that this is the most consequential acquisition that Qualcomm has made since Flarion (bought for 4G), as it gave it a launchpad from which to launch an M-series competitor for Windows laptops.
• It also offers the possibility to roll the processor design across all of its other segments with smartphones coming in 2024 and I suspect automotive and cloud soon to follow.
• Qualcomm was always going to find this much harder than Apple as it lacks the vertical integration that Apple has which made it much easier to ensure that hardware and software were optimised to work together.
• Furthermore, it also has to deal with the huge diversity of hardware specifications and range of potential components all of which Intel has been dealing with for decades.
• The fruits of this development were launched at Snapdragon Summit in 2023 where, if Qualcomm’s numbers are accurate, the Snapdragon X Elite processor is only marginally behind the M3 Max chip (see here).
• Qualcomm does not have to deal with Apple in laptop processors and so the benchmarks that really matter are against x86 processors and here the X Elite is on strong ground.
• I am expecting that Qualcomm will launch the Snapdragon X chips pretty soon and it looks like there will be 8 or more SKUs.
• Four of these would be for high-end laptops powered by the X Elite and 4 for mid-range powered by X Plus (see here).
• This is an early indication of how seriously Qualcomm is going after this market and every conversation I have had indicates that the emulation is working better than Rosetta Stone (Apple’s) and at lower power consumption.
• This is crucial as it is this that will ensure that all Windows apps run on arm-based Windows machines and do so in a way that user does not notice.
• Only if these laptops offer a seamless experience will the users adopt them, and we will begin to see the first devices launched at Computex in June this year.
• The effort that is being put into this is orders of magnitude greater than last time this was tried which I think gives it a much greater chance of success.
• In the short to medium term, this could be much more consequential for Qualcomm than AI at the edge in smartphones, as it is a large market with high processor ASPs where Qualcomm has almost 0% share.
• Furthermore, with Apple not competing in merchant silicon and Arm’s version of this (Blackhawk) probably 6 months away, Qualcomm could have 12 to 18 months of only having to compete against x86 in the laptop CPU market.
• If Microsoft has done its job properly and the port to Arm works seamlessly, then there could be a material increase in Qualcomm’s revenues from taking share from Intel and AMD.
• AI at the edge makes a lot of sense and also offers plenty of upside but I think it is further away in terms of time and in many edge devices, Qualcomm is already present.
• However, while the technology media and glitterati continue to guzzle AI cool aid by the gallon, this opportunity is being completely ignored meaning that there could be some surprises in store later in 2024 and 2025.
• The easy money in Qualcomm has been made with the recent rally from $106 (where I picked it up) but when considering taking profits, I have concluded that there is more upside.
• The shares are no longer a bargain but at 15.2x 2024 and 13.8x 2025 PER (which will be lower if any of this new strategy pays off), I am very happy to hold on as I am sure there is more to come.