While the AI boom has driven DRAM prices up by over 560%, NAND Flash is entering a critical "silent squeeze." With Enterprise SSD demand outpacing supply and major producers like SK Hynix (000660.KS) already thought to be sold out for 2026, the market is bracing for double-digit price hikes.
NAND vs. DRAM: The AI Price Gap
Much of focus on memory prices has been  on DRAM as the obvious beneficiary of the AI boom.

• DRAM Price Increase: +561.1% (Average)
• NAND Price Increase: +63.6% (Average)

While a price increase of over 63% might seem extreme in a more normalized market, it pales in comparison to DRAM’s rise.  That said, NAND producers, typically the same group that produces DRAM seem to be signaling that they will be raising NAND prices again in 1Q.  This is despite the rise this year, as demand for NAND continues to grow and the fact that NAND products are experiencing another round of price increases as we write this note.
2026 NAND Price Forecast: The Supply Shortage
 We have seen predictions that NAND prices will rise again in 1Q by more than 10% (Current Forecast) and 20% to 30% for the full 2026 year, based on demand growth of 20% to 22% against capacity growth of 15% to 17%, with rumors that supplier SK Hynix has already sold out its 2026 production an indication that negotiations for those in need of volume in 2026 have already taken place. 

• Q1 2026 Forecast: Prices expected to rise >10%
• Full Year 2026 Forecast: Prices projected to increase 20% to 30%.
• Supply vs. Demand: Demand growth (20–22%) is projected to exceed capacity growth (15–17%).

Why Are NAND Producers Restricting Capacity?
Enterprise SSDs and data center storage continue to drive that NAND demand while producers are being very cautious about adding capacity.  We believe there are three reasons for that caution:

• The 2022 Price Collapse – In 2022 the price of 512Gb wafer prices fell from ~$4.00 to ~$1.40.  Inflation, the end of COVID demand, and considerable overproduction were responsible, and left producers selling product below cost.
• The Capital Allocation Choice – DRAM or NAND – Which of the two potential products would be more attractive to producers?  The obvious choice, at least currently, is DRAM as HBM is the crux of demand in the AI server space.  While NAND is getting more attractive as prices rise, if you have to choose, NAND comes in second.
• The Potential AI Bubble – Perhaps a bit more speculative, capacity expansion caution based on the possibility of an AI implosion always looms a potential factor when spending billions on new capacity.

These three factors alone are likely able to keep spending for NAND capacity below demand levels but adding that knowing you will likely sell out most, if not all of your 2026 capacity early in the year, might also contribute to the NAND capacity expansion hesitancy.
Top NAND Flash Producers by Sales (Q3 2025)
​

​Conclusion: The Silent Squeeze in Storage
While the semiconductor spotlight remains fixed on DRAM and the astronomical rise of HBM, NAND Flash is quietly entering a period of structural shortage that seems to guarantee upward price pressure through 2026. The market dynamics suggest that the current 63.6% price recovery is not a ceiling, but rather a baseline for the next fiscal year.
The governing force for 2026 will be producer discipline born from trauma and opportunity cost. The memory giants (Samsung, SK Hynix, Micron) are effectively "shorting" their own NAND supply. By prioritizing high-margin HBM production and refusing to expand NAND capacity due to the scars of the 2022 collapse and fears of an AI bubble, they have engineered a seller’s market.
With demand growth (20–22%) structurally outpacing capacity growth (15–17%), and major players like SK Hynix already reportedly sold out for the year, the leverage has shifted entirely to the manufacturers. For investors and enterprise buyers, the narrative is clear: NAND may not be the volatility "sprinter" that DRAM is in the AI era, but it has become the steady, high-yield climber. As data centers fill with AI processors, the critical need for high-performance storage (Enterprise SSDs) ensures that NAND producers will enjoy sustained pricing power and revenue growth well into 2026, free from the threat of imminent oversupply.

The Smartphone Battery Capacity War: Why mAh is the New Feature Battlefield
The New Rules of Smartphone Competition: Feature Trends vs. Raw Endurance
Smartphone competition is fierce.  There are between 10 and 15 major global smartphone brands but almost 150 regional, local, and niche brands that release at least one model each year..  Samsung’s overall marketing department has  over 23,000 employees and with smartphones being a major Samsung product, a large chunk of the company’s ~$10 billion advertising budget goes toward smartphones..  Apple (AAPL) has a smaller marketing department and a smaller number of products but still spends between $2b and $2.5 billion on advertising each year, so smaller brands compete by offering consumers features.  Over the years there have been ‘feature trends’ such a larger screens, multiple cameras, and most recently AI, but there is another more subtle ‘feature battle’ brewing that actually benefits consumers.
Smartphones are marvelous devices.  They allow us to communicate with almost anyone on the planet, they give us the opportunity to capture life in pictures, and they allow us to perform dozens of tasks that only a few years ago would require sitting in front of a computer monitor.  Of course there is a downside, a serious reduction in face-to-face interaction and a more couch potato society, but they have become an indispensable part of daily life. Mobile phones have one big limitation, they operate on batteries which have a limited operating period before they need to be recharged.  Just as lager screens on smartphones hit a wall (pocket size) at a bit over 6”, batteries are constrained by the size of the device and must compete for internal real estate with a wide variety of other components,  Battery technology continues to improve but what good is even the best smartphone when the battery is at 2%.
Battery Capacity Trends: Matching mAh to the Needs of AI and TBC
 Over the years, battery engineers have battled with circuit desiner5s to match the needs of smartphone hardware to battery capacity to keep consumers from having to charge their phones in the middle of a workday or at a special event. Now that battery technology is improving, battery size, measured in mAh continues to grow and seems to be the next ‘feature battlefield’ for smartphone brands.  Some of that battery competition is not competition at all, but a necessity as Ai applications are a new source of power draw that must be accounted for  in order to maintain TBC (Time Before Charge).  But on top of that there seems to be a bit of one-upmanship between brands that exceeds raw power needs.  Here is how batteries have grown over the years using two high-end smartphones as examples:

Aggressive Capacity Growth: Xiaomi, OnePlus, and Google Pixel Trends
 
Understandably both battery capacity and TBC are increasing, but when we start to look outside of these two very well known brands and models, things become more obvious.  Here is a similar comparison between One Plus (pvt) Google (GOOG) Pixel, and Xiaomi (1810.HK):
​

​The Next Frontier: Will 10,000 mAh Phone Batteries Reset the Industry Standard?
But here is the kicker.  Honor (pvt) has under development a phone with a battery capacity of 10,000 mAh.  The company has confirmed one model and has been rumored to be working on a second.  This would be a substantial increase over the current model (7,500) and an even larger increase over the competition, leading one to believe that the war has begun and Chinese brands are leading the battery charge.  One Plus has a similar project under development, although the timing looks to be greater than one year for such a release.
Conclusion – The New Feature Battlefield is Endurance
The detailed comparison of battery capacity (mAh) and Time Before Charge (TBC) across major smartphone brands confirms a critical shift in the competitive landscape: the battle for device differentiation has moved from aesthetics and camera resolution to raw endurance. While market leaders like Apple and Samsung exhibit incremental, calculated growth in capacity (driven primarily by compensating for new power draws like AI features), challenger brands, particularly those from China like OnePlus and Xiaomi, are clearly engaging in a competitive "one-upmanship."
This analysis shows that the necessity of powering complex AI processes is merging with an aggressive market strategy, culminating in unprecedented capacity targets like Honor’s rumored 10,000 mAh phone battery. This movement by challengers to drastically increase mAh and TBC is not merely a feature trend; it is a fundamental disruption aimed at carving out market share by solving the user's single greatest pain point: battery anxiety. As AI integration accelerates, consumer expectations for all-day and multi-day battery life will solidify, ensuring that Time Before Charge becomes the next definitive benchmark of flagship quality. Brands that fail to aggressively pursue superior battery solutions risk being left behind in this new era of mobile endurance.
 

We thought, after weeks of price gains that pushed prices above trendlines once again,  that DRAM pricing might take a breather last week.  We were wrong, in fact the price of DDR4[1] continued to rise making last week the fourth week in a row with price increases over 10% and pushing DDR4 to its high point for the year.    Little has changed fundamentally as the broad switch by DRAM producers from DDR4 to DDR5 continues, along with demand for HBM, which is capturing capacity that might still be producing DDR4, leaving the market tight.  It is expected that DDR4 prices will continue to rise into 2026.  The current spot price of $17.68 is the high for this year, up from $2.93, reached on 3/5/25, for a gain of over 500%.


[1] DDR4 16Gb 2Gx8
 

It is no surprise that DRAM prices are rising as Ai demand continues to expand.  We have already noted the actual and planned DRAM price increases from producers, but there is nothing better than a chart to show how the trend toward another round of DRAM price hikes is already affecting DRAM spot prices, and eventually CE prices.  Without any more mishegoss, here’s the current spot DRAM price chart:

Component lead times are among the best indicators of the balance between supply and demand and present an undeniable picture of how those trends develop.  As an adjunct to our note on component pricing from last week, we point to the data on automotive and non-automotive lead times presented by TechInsights which shows how automotive component lead times, that were so long during the COVID epidemic as to halt automobile production, reached ~15 weeks last year and have continued to slowly decline during 2Q to 12 weeks, after a small rise (12.9 weeks) in 1Q.  Non-automotive component lead times rose a bit in 2Q, from 12.7 weeks in 1Q to 13.2 weeks but remain above the 12 week level.  In 2Q every individual automotive component group saw declining lead times, with RF Semiconductors and desecrates showing the least declines (2% to 3%) and power MOSFETS and clock timing semis showing the greatest, between 12% and 13%.
New vehicle sales are the primary driver for automotive semiconductors, however this is a global market, meaning a pickup or decline in US new car sales is not the only driver of the automotive semiconductor market as sales in other markets, such as China (#1 position in units), Japan, India, and Germany can offset trends in the US market.  However, the automotive component debacle during the COVID crisis has caused component buyers to change their philosophy from JIT (Just-in-time) during the pre-COVID era to JIC (Just-in-case), a more conservative philosophy that leads to higher inventory levels to ensure component availability regardless of circumstances.  These higher inventory levels have led to the shorter lead times seen in the last 18 months and continues to keep lead times from expanding drastically during short-term demand cycles.

All in, automotive components are a special class of electronic devices that have unique applications in vehicles, especially hybrid or all-electric vehicles.  Components like SiC MOSFETs, DC-to-DC converters, or MCUs are among the highest value components outside of the battery pack and motor, and might have longer lead times (typically closer to 20 weeks or more) than more generic automotive components, but the general tone of lead times for automotive components has changed as buyers and supply chain managers still have a strong impression of the COVID years to keep them focused on JIC, and while automotive component lead times still react to global and local automotive sales trends, the overall lower lead times reflects that lasting impression. 

Surveys are inherently flawed.  While survey results are presented as statistical data, survey questions are biased, and survey responses are more so in that they reflect what the respondent says which is not necessarily what the respondent actually does.  That said, in many cases we have few, if any, alternatives, so we use surveys with a grain of salt and try to integrate them with hard data.  In the case of electronic components, the ECIA (Electronic Components Industry Association) is quite open about the fact that its survey data is based on sentiment, which can be affected by very short-term influences, such as news events, or long-term influences, such as elections, and as all of the respondents participate in the component industry in some way, there is an overarching bias toward an optimistic view of the industry that pays their mortgage. 
The ECIA component survey queries three groups for its responses; component manufacturers, distributors, and manufacturer representatives from 6 component categories, 6 semiconductor categories, and 8 end markets to get a broad view of the component industry and the prevailing sentiment for the month, quarter and year, along with a forecast of the upcoming month, all generating a compiled index.  The index value represents the overall sentiment of these groups, with a reading below 100 indicating a negative growth outlook for the industry and over 100 a positive one, as shown in Figure 1 and Figure 2. 
After a dismal end to 2024, component sentiment has been positive for most of this year, although the ‘fear of tariffs’ turned sentiment negative toward the end of 1Q and into 2Q.  Since then overall sentiment has been quite positive, between 10% and 20% above the growth line.  As can be seen in Figure 2, the July sentiment forecast (dotted lines) was quite optimistic, with expectations of a move from 121.6 (actual) to an additional almost 5% boost for August (127.5 forecast), indicating a continuation of the optimistic spirit seen in much of 3Q.  However it was not to be, with the actual August sentiment reading not up 5% but down 7%, a surprising turnaround, with all three product categories declining similar amounts. 
While the ECIA did not give a reason for the large misstep, we surmise it was again the tariff situation and the prospects of extended negotiations to reduce recent high tariff rates.  The most optimistic group were manufacturers’ representatives, with distributors seeing some improvement based on component demand conditions, while manufacturers, who are most directly affected by import tariffs, were pessimistic.  Similarly at the top of the optimism scale were participants from the Industrial & Avionics segment (includes military & space) and the bottom of the list were those from the consumer electronics industry, the only end market group with a sentiment index below 100, neither of which seem out of line with one might expect given the political and trade agenda.
But have no fear, as it can be seen in Figure 1 that overall optimism remains the watchword for September, with all product groups spiking upward, some reaching new highs for the year.  Again, after a big miss last month, one would expect a bit more caution, but that does not seem to be the case, either because of a sense of strong demand from customers at the end of 3Q or just end-of-the-year optimism.  Still, on a more general basis, optimism remains for 3Q and 4Q, although we expect that a second month of less than expected survey results might dampen that enthusiasm.  Another 4 weeks and we will know…

While we track memory prices, we don’t often comment on spot prices unless they are doing something unusual and for the last few weeks DRAM prices have been heating up, with this week seeing the greatest w/w gain of the year in most DRAM categories.  This has pushed DDR4 prices above DDR5 prices, typically something that does not often happen in the daily course of events.  DDR4 supply is tight but much of the panic is coming from the lack of a definitive trade agreement details with China that might settle potential semiconductor and memory trade issues.  The steep part of the increase has been in a relatively short timeframe but should it extend it will add to the cost pressure CE companies are feeling before they start worrying about tariff issues…

While the world watches the daily Trump tariff circus erode any near-term chance of stability in the housing market, the CE space and even the avocado market, there is another problem afoot that most people have yet to see.   Cobalt is a key component of airplane gas turbine engines, and it is what makes blue glass blue, but its most common use is in the production of the anode in batteries, typically those used in mobile devices but more recently in electric car batteries.  China is the largest consumer of cobalt (~87%) and on a general basis EVs consume ~40% of the global supply, but what makes cobalt unique is that almost 75% of all cobalt produced globally comes from one country, and its not the US or China.
The Democratic Republic of the Congo produces ~74% of the world’s supply of cobalt and has ~54% of the world’s reserves of the metal, and while Australia produces between 2% and 3.5% of global supply, it is the only other country with more than 4% or 5% of global reserves (~15.5%).  This means that the DRC essentially controls the world’s supply of cobalt.  This is not a new occurrence, but something has changed recently that makes Trump’s tariff threats look mealy-mouthed in comparison.  On February 22 of last month the government of the DRC decided to halt all cobalt exports for four months.
The DRC statement indicated that the ban was put into effect “to regulate supply on the international market which is faced with a supply glut”, a reference to China’s CMOC Group (603993.CH), the globes largest miner of cobalt, more than doubling production last year, amid slowing demand.  The reaction was quick, as can be seen in the long-term and short-term charts below, with prices rising every day since the ban was announced.  That said, the long-term chart also indicates that the material is down severely from its peak in 2022, hitting lows last year not seen in over a decade.
The DRC has indicated that it will revisit the ban after three months to decide further action, but with the country holding more than half of the world’s cobalt reserves, it is expected that stockpiling has already begun.  Hopefully you bought that Tesla last year and won’t need another battery for 15 – 20 years unless it gets damaged.   Wait, do I smell something burning?

Smartphones are incredible feats of technology, squeezing innumerable parts into a rectangle typically under 20 in2 and ~ 1/3 of an inch thick.  Most smartphone owners rarely see what is on the inside of their phones, making purchase decisions (hopefully) on matching specifications to their use profile.  But inside those rectangles, packed in like sardines, are literally hundreds of ICs and other components, along with a battery, display, and assorted cameras. 
There are those that relish the thought of purchasing smartphones and then taking them apart, piece by piece, in order to quantify structure and cost.  Such a group is TechInsights, who are known for their detailed teardowns of various CE devices.  They have been kind enough to afford us a detailed look at one of their smartphone teardowns, which we summarize below.
The phone being disassembled here is the Sony (SNE) Xperia 1V, a device released in July of 2023.   Sony is not a major smartphone brand but is known for the high quality of their phones, so the example below should be a guide as to what to look for in a high-end smartphone.  We note that when the Xperia 1V was released, it sold for $1,399.  The phone weighs 188 grams, runs on Android, and has a 6.48” OLED display, along with four cameras, and runs on a Qualcomm (QCOM) Snapdragon 8 Gen 2 processor.
While the greatest share of the BOM is the broad category of integrated circuits (45.7%), the camera subsystem captures 23.7%, due to the fact that it covers 4 cameras and associated electronics, lenses, etc.  The display subsystem, which is a single 6.48” OLED display and a touchscreen, along with a polarizer and cover glass (total of 70 components), is next at 7.5%, followed by non-electronic parts (frame, etc.) at 7.2%.  More relevant to the investment community would be the breakdown of the total component types and the IC category on a branded basis.  As can be seen in the table below, the IC category carries the largest cost share by a large margin, putting significant weight on the brand share shown in the table that follows.  

Trackers have always been a part of history, with the tracking of animal food sources playing a pivotal part of human existence, but as technology has developed, particularly after the Russian satellite Sputnik was launched in October 1957, scientists and engineers realized that they could calculate location based on satellite location, and eventually launching atomic clocks into space that laid the groundwork for GPS satellite systems that are used today.  The global GPS system relies on satellites that are placed precisely 11,000 miles above the earth and therefore orbit once every 12 hours, but there are a number of global  satellite systems beside GPS (run by the US Department of Defense), such as GLONASS (run by the Russian Federal Space Agency), Galileo (run by a number of EU countries), Beidou (run by the Chinese government), IRNSS (run by the Indian government), and QZSS (run by the Japanese government).  Many of these systems have both military and civilian users and have become essential to the military as they are the basis for many targeting systems, while in civilian use, they tend to be used for geo-location and asset tracking.
As technology continued to develop, particularly with the popularity of smartphones, other tracking systems have been developed, although there are many GPS tracking devices and systems available to both individuals and commercial users for a variety of uses and prices, ranging from ~$30 for a 2.3” tracking disc and software, to larger and more powerful devices that can be used to track assets such as cars and shipping containers.  As these devices are powered internally, they have a finite lifespan, which runs from 1-2 weeks for a device that is always on and sending notifications, to 6 months for those that operate in low-power mode, with more limited communication ability.
A small company, Tile (LIFX) had been producing consumer tracking devices using Blue-tooth low-energy tracking since 2013, which had a range of ~100’ but included a system where any Tile user who was within 100’ of a lost Tile device would trigger an anonymous message to the owner, giving the location without the non-owner knowing (known as “crowd GPS”).  The idea was to attach a Tile tracker to common devices, such as keys or a TV remote, to make sure they were always able to be found, however, in 2019 it became apparent to developers that Apple (AAPL) had included in its iOS 13 release, a number of references to a product then known as B389, which fit with rumors that Apple had been developing a personal tracking device that would operate under iOS and use Apple’s in-house U1 chip that appeared in the iPhone 11. 
The Apple system used a different technology called UWB or Ultra-Wideband, which differs from standard radio transmission and Bluetooth, both of which use narrow frequency bands and transmit data continuously.  UWB broadcasts a signal that is pulsed but occupies a large frequency band, allowing it to be more powerful and carry more information without interfering with other devices.  This increased power is spread across a wide frequency band and appears as noise to other receiving devices but allows the system greater range as shown in the table below.

While other major CE brands have come out with their own UWB trackers, and a number of smaller companies continue to supply devices,  Apple is so ubiquitous that we expect they dominate the space, although we are hard pressed to find accurate shipment numbers for the entire UWB tracker space or for Apple’s AirTags.  Most estimates center around 20m units in 2021 and an expected 35m units last year, but there is little hard data that can be verified.  That said, there seems to be a resurgence in the idea that Google (GOOG) will be joining the UWB tracker race with its own tracker. 
The project, supposedly titled “Grogu” or G10, after the character in “The Mandalorian” TV series, is part of the company’s “Find My Device” initiative.   As Google purchased Nest (pvt) in 2014, the idea of a more connected home is essential to Nest’s success and in a 2022 Android update, Google included code that made obvious the company’s intention to popularize UWB by giving Android developers UWB hooks into the OS.  That said, we suspect the initial UWB focus was to enable connectivity between Nest and Google devices, such as to allowing a Google smartphone used as a music source, to move from one room to another, with speakers in each room able to instantly recognize the source and seamlessly continuing to play the music. 
We note also that the Google Pixel 7Pro (10/22) and the Pixel 6 Pro (10/21) smartphones are both UWB enabled, and while we have no real timetable for Google’s entry into the UWB tag market, the logic seems apparent.  With Android supporting ~70% of the of the smartphone world, Google has a more reasonable chance of encroaching on Apple’s personal tracker dominance than most others, and while the iPhone network gives life to Apple’s AirTag abilities, if Google releases its own tracking device, it will eventually have access to all Android phones that have UWB capabilities.  It is a bit of a chicken and egg scenario, but Google’s Android share gives them a good shot going forward.