​You might notice that today’s note format is a bit different.  We are trying out Ai software that help content providers optimize their work for SEO and GEO.  Most of the ‘suggestions’ are formatting changes (emphasis, restructuring tables, etc.) but along with those also come some suggestions about the copy itself, and we find modifying what we write about to make it more appealing to Google (GOOG) bots a bit disconcerting.  That said, we will continue the experiment for a few days to see if the bots take notice.  Thus far it seems that instead of providing noteworthy and readable content, the software encourages one to ‘shout’ rather than speak, but it is only day 1 and we don’t have an end game quite yet.
 

Staging is a common real estate practice and gets progressively more elaborate as the price of the real estate increases, making staging a necessary cost for high-end real estate but a costly perk for mid-tier apartments or homes.  Base price quotes usually start around $5,000 for a package including  item move-in and move-out, while lower priced ($2,000 - $3,500) might not include the moving costs which are outsourced.  In some cases realtors will front the cost of a staging, but will reimburse themselves from the sale proceeds, so the value of the real estate is an important determinant in making a staging decision.  Or it was…
It seems that AI is taking over that job and is causing some consternation as it can present a very different image from what the apartment actually looks like, although in a very competitive real estate market, drab photos are easily overlooked and well staged photos capture considerable attention.  Retouching house or apartment photos was considered gauche, and it took time and cost money, but now almost anything can be fixed by AI especially applications specifically designed to virtualize furnishings.  You can even choose your furnishing ‘style’ and with a few clicks you can make a messy ‘lived-in’ apartment or house into a showplace, without anyone asking any questions. 
What happens when you get to the apartment for a showing?  The agent can say “It just came on the market and I haven’t been here myself yet, but I have a few others I can show you”, or “Oh, that one isn’t available anymore, but I’ve got others”, and while there might be some trust lost, the agent’s traffic level is increased, and the cost is almost nothing to the agent or the seller.  Renters and buyers should be aware that anything that looks perfect is probably not real and should visit potential apartments or houses a number of times to get a true picture of what it really looks like and not trust images without a few reality checks.
To show how easy it is to do this, even at the most basic level, we took it upon ourselves to create an image of a New York City apartment living room, as if it had been staged for a showing:

​

 Note the flowers, the trees outside the window, the antique rug and the NY skyline view…

As we move back a step, one notes that the tree and the NY skyline view have been replaced with the original view of a typical NY airshaft and the bright light entering the room is somewhat dimmed..

​

Another click and various articles of clothing left during a rush to get to work are now visible.  Notice also that the flowers are gone.

Another click and the dirty dishes from last night’s dinner in front of the TV appear and the couch pillows are showing signs of a early evening nap.

This is what the apartment living room actually looked like (not ours), needing a few thousand dollars’ worth of clean-up and staging, but in less than 15 minutes we were able to make this disaster area into an appealing apartment that would generate lots of views and potential clients.  Our concern does not extend to staging company employees losing their jobs as there will always be a need for staging in the real estate market, but driving traffic by completely falsifying images seems to be catching on.  The image below is a real ‘touch-up’ out of China, where it is said that graduates looking for their first apartments are facing these modified images, only to be sorely disappointed when they actually visit the site and find it filled with the previous renter’s items and less than ideal appliances or damaged floors that had been changed in the listing images.
Caveat emptor!

We give Corning (GLW) credit for being in the right place at the right time in terms of their optical business, catching the AI data center wave with ultra-high density connectors and fiber.  While we hope that this leverage continues, we are still on the fence as to whether data center growth will continue at this rate for the foreseeable future.  That said, we go back to the two businesses that are the reasons we follow Corning, their display glass business and their specialty materials business.  The display business has survived a number of precarious moments, such as the fear that flexible OLED would cut into the company’s LCD glass substrate business, or the end of LCD display production by the two major panel producers in South Korea, Samsung Display (pvt) and LG Display (LPL), so we spend some time looking at these segments, letting AI play out a bit more before jumping on or off the data center bandwagon.
Figure 1, Figure 2, and Figure 3 show Corning’s display division sales on a long-term basis, and segment margins on both a long-term and 5-year basis.  The charts show that 2022 was a correction year for much of the CE space after the surge in demand during the COVID pandemic and the inventory build that fed such demand, Corning’s display segment recovered in 2023 and continues to follow a positive trendline.  During 2023 the company announced a substrate glass price increase (~20%), the first since mid-2021, helping to bring prices in-line with rising production costs due to inflation.  This has helped to bring display segment margins back to more normal levels after hitting 5-year lows in 3Q 2022.  Since then display sales have been relatively steady as the price increase held and inventory levels remain normal.
Display margins have also recovered, to the mid 20’s and more recently the upper 20’s, with every quarter this year at least 2% higher than the 5-year average quarterly margin.  We expect margins to remain under 30%, but as Corning’s highest margin business, it is critical that display margins remain in the mid to upper 20% range.  The company projects that display margins will be at least 25% for the full year on between $900m and $950m in sales but indicated that they expect sales to come in at the high end of that range, likely pushing margins above 26% once again. 
On a longer-term basis we see the display segment maintaining sales in the $3.7 to $3.9b range and margins between 26% and 27%.  While we expect substantial growth will be difficult as the display space itself remains in stasis, as long as Corning is able to maintain consistent volumes the display business should continue to generate substantial margins.  We note that Corning management also noted that as the only US display glass supplier, those panel producers looking to increase their US ‘content’ to offset trade restrictions have only Corning as an alternative.
The Specialty materials segment at Corning saw a strong quarter with sales setting a new company record, which we suspect might include some additional Apple (AAPL) business, apropos of the announcement made in August in which Apple committed $2.5 billion to Corning’s Harrodsburg production plant, making it the exclusive supplier of cover glass for all iPhone and Apple Watch products, part of Apple’s commitment to invest $600 billion in US manufacturing over the next four years.  That said, as OEM Gorilla Glass is refined outside of Corning[1] sales of Gorilla Glass do not always correlate to smartphone sales.
Corning’s Specialty Materials segment saw the same sales and margin contraction that affected the display space in 2022, but about a year later as smartphone shipments continued while other CE products slowed.  Specialty materials’ margins have been improving since, currently reaching 18.2% on record sales, which we believe would typically be at or near the high end of the long-term range.  However we expect that the Apple deal might give some room for margin expansion as Corning not only supplies cover glass for the iPhone and Apple Watch but is also the finisher  for Apple.  This should lead to some margin improvement over time as Apple moves all of its cover glass production to Harrodsburg and that facility no longer produces any raw glass for OEMs.
While we are certainly cognizant of the effect AI and data center build outs are having and will have on Corning’s sales and profitability, we are less sure about the long-term sustainability of that growth and look to Corning’s other CE related businesses for steady sales and profitability.  While growth might be a struggle for the display segment, it is a reliable source of sales and income for the company, and as the Trump administration will likely continue to hammer on CE brands to source from US companies, Corning has the competitive upper hand for the next three+ years.  Specialty materials carries a bit more risk in that it is focused primarily on the smartphone segment, but the concept of using more glass per device seems to be keeping demand growing despite the ups and downs of the smartphone market.


[1] The ion exchange strengthening process is done at the OED level

Ai is revolutionizing the world, and many see it as an invention that rivals fire in terms of its long-tern effect on society.  But there is another, slightly less seismic and certainly a bit less profound perception of Ai, it’s use in advertising and product development.  We are not talking about how AI might be changing the way advertising agencies produce copy, or how the average consumer receives ads.  We mean the concept held by many that almost every product would benefit from AI and that it should be included in almost everything we have contact with in our daily lives.  This seems to have become a common theme across almost all product categories with “AI” emblazoned across product literature and “Machine Learning” a common theme in descriptive text, but what is AI actually doing in products and is something worth paying for or just a gimmick?
 
Oral-B, part of Procter & Gamble (PG) household consumables business model, sells toothbrushes, toothpaste, dental floss and mouthwash.  Oral-B manual toothbrushes sell for a little as $0.99 while the Oral-B IO series 10 rechargeable Electric Toothbrush (AI) sells for ~$400.  Tooth care is important, as we get one set of teeth that are supposed to last for  lifetime, but do we need an AI toothbrush?  This electronic marvel has a broad feature set that outshines your ‘generic’ electronic (non-AI) toothbrush by doing the following:
 

• The toothbrush can track 16 (!) zones in your mouth to ensure complete brushing coverage.  The standard model has a 30 second timer that tells you to move to the next (of 4) zones.
• Via the Oral-B app or the iO “Sense Charger”, the toothbrush will show you a live map of your mouth, highlighting missed spots as you brush.  The standard model forces you to rely on your own sense of coverage.
• It has a pressure sensor and three indicator lights that specifies how much pressure is being used when brushing and slows down if you are brushing too hard.  The standard model only has a red light to indicate when you are brushing too hard.
• It has 7 cleaning modes (Daily Clean, Sensitive, Super Sensitive, Intense Clean, Whitening, Gum Care, and Tongue Clean).  The standard model can have anywhere from 1 cleaning mode to 5.
• The AI model has a magnetic drive (nothing to do with AI) that allows the oscillating motion to be combined with micro-vibrations.  The standard model relies on pulsations.

So you can leave your teeth in the hands of the Oral-B AI that can coach you every time you brush your teeth using a personalized, guided training experience, or you can go for the dumb electric toothbrush with its fewer features and only one LED.  Of course, sticking with the “I can brush my own teeth” model will cost you between $120 and $160 (Series 6) while the AI model will cost you ~$400, or, if you are really antediluvian, you could use a regular toothbrush.  For the amount that the AI model costs you could buy a regular toothbrush four times each year for the rest of your life.  Is Ai necessary here?
 
Here's one that gets more positive press, the Ai thermostat.  Manufacturers tout using an AI (not just ‘smart’, but ‘AI’) thermostat will generate10% to 12% savings on heating bills and up to 15% on A/C, using a $900/year HVAC base.  Similar to the toothbrush scenario, there are two alternatives to the AI thermostat, the first being an old style thermostat that you set to a particular temperature and it maintains that temperature forever.  The other alternative is a digital thermostat that allows the user to program temperature settings based on the time of day and the day of the week.  The Ai thermostat offers the following features:
 

• Algorithms that learn your routines (home, away, day, night, etc.) built on occupancy sensors or geofencing based from your smartphone’s location and movements (GPS).  Digital thermostats require the user to program in scheduled temperature changes, and generic thermostats stay at the same temperature regardless of the time or day.
• The AI system measures how long it takes to heat or cool the home and uses that to anticipate  the time it takes to bring the home to the desired temperature.  Digital thermostats are triggered by time and temperature, so again they rely on the user having set those parameters.
• AI thermostats use temperature averaging.  Wireless sensors are placed in various rooms, and the system adjusts the overall temperature based on an average of all  of the rooms.  Unless the system is able to be adjusted on a room-by-room basis (rarely unless using mini-splits) averaging would make little difference for rooms that are north facing or South facing.  Digital thermostats do not make any such calculations and heat each room based on the temperature at the thermostat point, making placement important.
• The one part of AI thermostats that makes the most sense is the ability to use sensors to note modifications to the normal schedule.  Should one take a half-day and arrive home at 3PM instead of 6PM, the system will sense that someone is in the house and adjust the temperature by pulling forward the typical pattern.  Should the same scenario occur with the digital thermostat, one would have to manually raise the temperature.

A single temperature (non-programmable) thermostat costs between $20 and $50, while a digital programmable (7-day) thermostat costs between $40 and $100.  An AI thermostat, including sensors,  costs between $130 and $400+  Whether AI in this scenario has value depends on the user.  A well programmed digital thermostat should be able to maintain year-round comfort for the homeowner, but it takes some experimenting and some refinements.  It cannot account for ‘special situations’, which would require the user to actually touch the thermostat to bring the temperature in line and the overall savings might be less than those from the AI models. 
 
There is some nuance here, as utilities tend to classify thermostats that are not AI but can be programmed via Wi-Fi as ‘smart devices, giving substantial rebates to those devices, so AI is not necessary to qualify but wi-Fi programmability is.  The average rebate in our neck of the woods is $85, which can bring down the cost of a Wi-Fi programmable thermostat to almost nothing, although our utility reserves the right to change the setting on your thermostat during the summer up to 10 times for a maximum of 4 hours to reduce stress on the grid.  All in, AI thermostats are likely able to save a user ~$100/year in ideal circumstances, against a non-programmable or poorly programmed thermostat.  A well programmed thermostat for a user with a steady routine would likely save a similar amount.
 
One last one, and one a bit more controversial.  AI TVs.  AI TVs are peddled as having the ability to analyze content, down to the pixel level, and dynamically enhance content and they do just that.  After training on thousands of images, they learn to break down video on a frame-by-frame basis and isolate objects, backgrounds, faces, text, and a variety of other objects, and find ways to enhance each frame.  This is in contrast to non-AI video processors that are based on non-adaptive algorithms that do not break down frames into component parts but apply a ‘fix’ to each frame based on its overall characteristics..
 
This comes in handy when the set is upscaling, taking low resolution content and expanding it to fit a high-resolution display.  The object is to fill in those ‘new’ pixels that were created and AI systems can refer back to their high-res training data to see what those missing pixels might look like.  It is also used to insert or remove frames to smooth motion by predicting how moving objects can best be displayed.
 
We don’t disagree that AI is very adept at creating ‘fill in the gap’ pixels, better than the standard method of interpolating neighboring pixel information, but there are two problems associated with AI video processing.  First, while TV engineers have display tools that go far beyond human perception, the average person is not going to see much of a change between an AI processor and a non-AI processor.  In some circumstances it might be noticeable, but most likely not on a day-to-day basis.  Second, there is an AI premium.  Depending on the price tier, the cost of adding an AI video processor can be anywhere from ~$300 to $1,000+ vs. a non-AI, but very capable processor.  Mid-tier sets would see a lower cost spread, and low-end TVs would not be AI enabled in most cases.  We know we will get some feedback from some of our readers and clients who are close to or in the TV industry, but we try to look at AI from a user’s standpoint, and we stick to our guns.
 
There are many more AI enhanced products that should either not be labeled AI or gain little or nothing from its inclusion, other than a premium price, and the 2026 Consumer Electronics show in early January will abound with ‘AI pogo-sticks’, ‘AI Hair Dryers’, and the ever popular ‘AI refrigerator’.  There will always be some consumers who will opt for such madness, but the average Joe does not need another device nagging him to eat his vegetables or instructing her to let the hair dryer cool before using again, and they certainly don’t want the cost of AI to be added to those products unless it can actually create value, rather than the perceived value that brands tend to emphasize.

We have mentioned the recent poor results of the Chinese National Day festival and how much of that weak sales performance was based on a reduction in subsidy funding from both the Chinese State and provincial governments.  While pinning down the exact cause is an impossibility, we do note that the Chinese spirit is not easily broken (at least publicly), and the Chinese CE space is filled with what it believes to be pointers to another cycle of TV set growth.  This is visibly evident in the TV space where even a 10.4% y/y decline in TV set shipments in 3Q (down 6.6% q/q) and comments in the trade press like “the third quarter of this year was the only third quarter in at least the past five years to experience a month-on-month decline in the Chinese TV market,” still lead to how TCL (000100.CH) and HiSense (600060.CH) are leading the world in the RGB Mini-LED market and eventually question as to whether 11/11 and the domestic TV market for the full year are ‘really that important’.   Only a few months back they were considered vitally important.
 
What it really comes down to is that while government support to stimulate consumer spending for CE is an effective short-term tool for boosting performance, as soon as the faucet gets turned off, the industry is back where it started, with little or no progress as to finding a solution for the basic problem, that of a lack of TV innovation.  Television grew up as a communal viewing platform where people could congregate and watch a program at a particular time on a particular night.  VCRs, streaming, and smartphones ended that practice years ago, but the TV industry has never moved on, still trying to stem the tide that pulls viewers to more convenient platforms.  TV size growth stemmed the tide for a while as almost life-size screens become more normal, but there is still something missing.
 
Just like the Chinese optimistic view of a poor situation, the Tv space is hoping for a savior.  A few years back it was OLED, and for a short moment it was Micro-LED, and now it’s AI, but little has really changed and there is little on the horizon to propel the TV space forward.  Technology is not a problem, as new display technologies are under development and quality continues to improve, but that does not solve the problem of how to get some of the inherent excitement that comes from a group of people watching something on a screen.  Yes, today hundreds or thousands can be watching the same movie or video on monitors, phones, tablets or other displays, but they are not in the same room and there is little of the shared excitement that came from watching the Beatles on Ed Sullivan for the first time, Nixon’s farewell address, or the dismantling of the Berlin Wall.  That shared excitement is missing as part of the TV experience, and without  that even large TVs are just Time Square advertising.
 
We are just as confused as how to solve this TV set dilemma as executives, engineers, and designers in the TV set space, but a solution is necessary if TV sets will ever again become a growth engine for CE.  There will always be ups and downs, but we expect there is little on the technology horizon that will change what we expect is a no-growth scenario.  How do you get people to sit down with each other and enjoy something communally?  That’s a difficult question in a world that is increasingly private and isolated; a world where conflict is increasing faceless and discussions typically end in name calling.  We don’t have an answer, but we believe whatever it turns out to be it will require that no smartphones be present.

​We have previously presented a bit of data concerning the Chinese National Holiday that began on September 25 and ended on October 5, primarily general data on the broad CE space.  Since then we were able to squeeze out some specifics on TVs during the holidays.  It is not much but it certainly gives some insight into how the diminishing subsidies are affecting various CE products on the mainland.

• During the holiday 1.15 million units (color TVs according the Chinese data, which we read to mean LCD TVs).  This was down 21.8% y/y and down 8.2% against 2023 volume.  Sales were ~5.3 billion, down 22.3% y/y and flat with 2023.  Some noteworthy points were:
• The trend toward large TV sizes slowed
• The share of 75” TVs, the China’s ‘average’ size category, declined by 2.3%
• Mini-LED TV penetration was over 30%
• More specifically for the on-line market, Mini-LED share was 31.8%, up 2.9% y/y and was 36.9% for the off-line market in China, up 5.8%

The tone of reports on the poor sales results for the holiday were still upbeat, despite the gloomy results, but the Chinese TV space has been defining its growth as the product of  price increases and larger average TV size growth.  As prices for Tv panels have been declining of late, the lack of TV average size growth is alarming.  While panel producers have been able to use utilization as a tool for balancing supply they have not had to grapple with slow size growth in addition to flat or declining panel prices.  Whether this is a new paradigm or just a short-term cycle is unknown, but in either case it does not bode well for 4Q results in the TV space.                                                                                                                                    

According to a number of sources, the recently released iPhone 17 series is selling better than the iPhone 16 series in its two largest markets, China and the US.  As such headlines are typically serving to garner clicks and contain little real information or data, we took things a bit further to try and flesh out what those headlines really mean.
Apple (AAPL) avoids the term MSRP (Manufacturers Suggested Retail Price) for legal liability reasons, referring to ‘retail price’, but retailers themselves often use MSRP to indicate ‘list price’ for the phones.  A comparison of list prices of iPhone models for the years 2015 – 2017 is shown below:

Given that the initial price for the iPhone 17 base model is $100 cheaper than its series 16 equivalent, it is difficult to evaluate the influence other factors (features, style improvements, etc.) might be having on early iPhone 17 sales, but given that both the US and China are the iPhone’s largest markets, we also compare the US base price to the Chinese base price below:

As can be seen, the Chinese MSRP prices are higher than those in the US, giving an advantage to US buyers, but both US and Chinese buyers are also offered a number of ‘discount’ options that can reduce the price of upgrading or switching carriers to as low as zero, however almost all carrier discounts still require full payment with a 24, 30, or 36 month bill credit repayment system that also typically requires signing up for a new (usually the high-priced plans) plan.  Apple and others offer trade-in discounts, in some cases regardless of the condition of the trade-in, although the values drop quickly when you trade in older models.
In China, retailers offer coupons (ostensibly for the 11/11 shopping holiday, although it is mid-October) and rebates that are specific to retailers and platforms.  T-Mall (BABA) offered coupons that lowered prices of some models by ¥300 ($42.13), while offered coupons up to ¥900 ($126.38) which lowered the prices of some models below the US MSRP, and almost all offered interest free installment pay plans.
So with all of this price chicanery, did the numbers for this year pane out?  On a combined basis over the first 10 days of availability the iPhone 17 series outsold the iPhone 16 series by 14%[1] y/y, with 8.8 million units sold versus 7.7 million last year.  Here’s how it breaks out between the US and China, along with model growth:


[1] Source: Counterpoint

We estimate that Apple sells between 72 and 90 million iPhones in the 4th calendar quarter of the year (see table below).  While this can be heavily weighted toward the current year’s models it also includes sales of previous year models, making current iPhone series estimates difficult and basing 4Q unit volume on a 14% y/y gain during the first 10 days does not account for the natural tapering of enthusiasm that occurs in each sales cycle.  While we would like to get a longer picture as to iPhone 17 series sales, especially in the US, we estimate a 4% to 5% improvement in total iPhone shipments in the 4th calendar quarter, or 80.7 million units.  Not a record, but certainly an improvement over last year, which was up 7% over the previous year’s models (iPhone 15) during the 1st 10 days of sales.  Rumors have been circulating that Apple has already cut the number of iPhone 17 Air units it is having assembled and increased the number of iPhone Basic units by ~5 million.  Whether that number will all appear in 4Q or be spread into 1Q or 2Q is unknown.

​All in, this is certainly positive news for Apple, but we hesitate to overstate the potential given the ongoing trade battle between the US and China.  Apple is certainly not hesitating to ‘kiss the ring’, making contributions to the WH Ballroom project, pledging a new $100 billion commitment to  US manufacturing, and announcing plans for a new AI hardware plant in Texas.  But staying on the good side of the President is a hit or miss proposition and companies can move in and out of favor quickly. And if sentiment sours further anti-American sentiment in China could flare.  It seems that Apple was able to maintain the excitement needed for a replacement cycle.  Now they have to keep that momentum going for the next year. 

Before we go further, September is almost an irrelevant month this year.  We know that shipments were up as brands added to inventory levels and panel producers pushed out as much product as possible.  We also know that there was a change in October.  As demand  slowed, brands pulled back orders, and panel producers lowered utilization in order to hold panel prices as steady as possible.  As noted above, that strategy has been working for many panel types, with only minor price degradation, but it has not been working as well for TV panels where brands are concerned that they might have built inventory level too high. This makes October the relevant month, where we expect to see lower shipments and lower sales, especially for those panel producers that specialize in TV set panels.  WE don’t expect a blood bath, but it will certainly be a more decisive month  than September, which was sort of a “What, me worry” month.
The metrics for the month (September) are:

• Shipments were up 4.1% m/m, up 5.0% y/y, but down 2.2% on a cumulative y/y basis.
• TV shipments were flat m/m while all other categories were up between 1.4% (Monitors) and 5.6% (tablets)
• All, including total shipments were up y/y, although monitor panels and total shipments were down on a cumulative y/y basis.
• Most large panel producers saw m/m sales gains, with LG Display (LPL) the strongest, and most small panel producers seeing sales flat or down m/m.

Display panel pricing, for the most part was flat for the month of October, with TV panel pricing the only category that saw movement (↓2.1% m/m), albeit within our forecast range. Our November forecast is almost identical to October, with a bit of an increase on TV panel downward pressure and less on mobile panel pricing.  We show both October pricing and forecast for each panel category below and also show October Y/Y and an additional metric that shows cumulative prices y/y, indicating how much better or worse the year is at this point relative to last year.

As we have noted previously, panel producers, particularly Chinese panel producers who control the LCD panel market, have been taking a different path this year.  Rather than letting panel prices fall to improve demand when buyers hold back orders, they are lowering utilization rates.  This has created a more stable pricing environment for IT panels (Monitors, Notebooks, Tablets) but has not worked as well for TV panels where TV brand buyers have faced considerable uncertainty concerning holiday demand.  Panel producers have lowered utilization rates temporarily in October but are hoping that brands will pull in some early 2026 orders before year in order to take advantage of lower TV panel prices.  Some of that optimism comes from the upcoming 11/11 holiday in China, although the  most recent holiday sparked little incremental demand, and the knowledge that the Winter Olympics will take place in early February.  However thus far brands have been holding out orders, assuming that TV panel prices will continue to fall.

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