A personal tale this month, since we decided to switch our home to a more self-sufficient and less carbon-intensive form of power. This is a topic that many people seem to be interested in, so I thought I’d share some experiences and, early days though it is, some numbers.

If you think climate change isn’t real and that burning coal is just fine because we’ve done it for longer than many countries have existed, then perhaps this journey is not for you. Given the seemingly endlessly rising energy costs fuelled by the clowns and fiends in global government, now felt like a good time for me to make this move.

Saving money in the short term wasn’t on the cards, given that it costs quite a bit to make this happen as a retro-fit on an existing house. It might make the house a bit more desirable if and when we come to sell it, but that again is not the reason to do it. Also, like any tech decision making, the end goal sometimes gets a bit lost in the process and you find yourself doing things just because it would be “better” (or just “because”).

The catalyst for us was the decision that we’re not going to be moving home for the foreseeable future. We decided, therefore, to fit the solar panels.

The business case for solar + battery

As soon as you start looking into this area, you’ll likely see lots of online ads punting the idea that solar panels can pay for themselves in no time, or even that there are schemes where they won’t cost you anything to install. As with all such things, your mileage may vary and there are plenty of operators who are maybe best avoided.

The first thing to think about is how much energy you are currently using, as measured in kilowatt-hours. As any skoolboy kno, this means if you had a device (like a kettle or a heater) which says it uses 2,000 watts, then in a whole hour of it working at full tilt, it would consume 2kWh. At least for UK users, you’ll see how much power you use by looking at the meter readings on your electricity bill.

If you could reduce your power usage by (say) £100 a month, it’s still going to take you some years for the project to pay for itself, but there is a certain degree of pleasure in being more self sufficient and it gives techie data nerds something to analyse.

The idea with photovoltaic (PV) solar panels is pretty simple – you generate power when the sun shines and either use it right away or store it for later in a big battery. In many respects, it doesn’t make sense to go for large solar installations unless you also have the battery since you’re probably not going to be using many kilowatt hours of electricity during the middle of a sunny summer day. If you don’t have a means to use, store or export the energy to the grid, then the panels do nothing – i.e. if your battery is full, your house is ticking over using only 100w and you don’t have a tariff that allows you to send surplus energy to the grid, it won’t matter if you have dozens of panels because they’ll only produce what can be consumed.

Our house isn’t 100% ideal for solar panels, as the rear roof faces south-east and there are some large trees that shade part of it until the sun is high in the sky. Still, modern PV panels can be tuned to deal with partial shade. At least we have a decent expanse of roof space, and that led to the first of many decision points – how many panels to fit? Advice was given by experts on the MoneySavingExpert forum, including one who actively gets involved in sense checking quotes and even negotiating with suppliers to get better deals.

The panels themselves are rated at 550w each, and the installer talked us into adding more than we initially intended  (2 rows of 9), just to fill the roof space. The idea being that a significant part of the cost of the project is the installation itself, and while you’re up there putting up 18 panels you might as well stick another row of 5 along the top. So that means we have 23 x 550w = 12.65kw in theoretical output. Of course, it never gets that high but there are plenty times when I’ve seen them produce over 10kw.

We were recommended to get a Tesla Powerwall 3 battery, partly because it contains everything in a pretty neat package, offers some additional capabilities like being able to power the house if there is a supply outage, and it gives a good usable capacity – it will store up to 13.5kWh of power. There are cheaper alternatives and some which give more access to be able to tweak and control what’s happening, but we thought we’d want the one that seems to offer the best mix of features, even if the company CEO is a lunatic.

As well as the mobile apps, if you have a Windows PC and use a Tesla Powerwall, you can monitor it with the free Powerwall Companion …

If you’ve a Home Assistant installation, you can get some basic info from the Powerwall on the local network, though Tesla shut down some of the functionality that was there in the Powerwall 2.

We were previously with an energy supplier who gave a very good rate compared to the mainstream ones, but they didn’t offer any form of export, so we decided to switch to Octopus Energy. With the Iranian conflict causing a tumultuous time in global energy markets just when the installation was going on, it took a while for Octopus to be able to get us on a good import/export tariff, which means we can sell excess energy at any time.

The tariff lets us import energy for a few hours overnight at nearly half price, but there is an evening peak for 3 hours where it costs nearly 1.5x the regular price. Similarly, exporting during peak hours pays a lot more than during the rest of the day. Even without solar panels generating anything, it would be possible to charge the battery up fully in the off peak time and export all of it during the peak hours; great though that sounds, it would only make about £1.50 each day, so it would still take about 15 years to cover the cost of the battery.

We also gave Octopus permission to manage the state of the battery – so they can decide to charge the battery automatically in the off peak period and keep its charge high during the day, so it can be discharged during the evening peak. They call this “Virtual Power Plant” and by using several thousand houses together, essentially give the equivalent of another power station. In principle, I could build Home Assistant automations to tweak how much charge to keep etc, but this way is just easier.

The costs

Total outlay for the installation of the solar panels and the Powerwall came to just under £14,000, which is quite a lot more than I’d initially thought; early investigations were on the assumption of having fewer panels and a smaller, cheaper battery. I got into the “sod it, do it properly” mode when planning started and we ended up with a bigger and better system.

When the Tesla system was installed, our existing EV charger (feeding a PHEV car with only about 11kWh battery) was kept “outside” the Tesla gateway – the idea being that you don’t want to plug the car in and it drain the stored energy in the Powerwall. In hindsight, I should have got them to do it the other way since the charger can be solar aware, and it would be possible to set up rules like only charge the car using excess solar power unless it’s the middle of the night and electricity is cheap. I might come back to this next year.

Since switching to Octopus in March, we’ve used about 3,200kWh of electricity from the grid, costing about £800 including the standing charge, though have exported nearly 3,800kWh back, earning us £585. So the net cost of electricity for 8 months is £215.

In comparison, for the same period in 2025, we used 4,700kWh electricity and paid £1,060. We also used LPG for heating, hot water and cooking – consuming around 6,000kWh of equivalent energy and costing about £500. So this summer, we’ve saved about £1,300 in electricity and LPG charges, even though we got a new PHEV car with a bigger battery than previously, and made another significant change – switched from LPG to installing an Air-Source Heat Pump.

Pump up the Jam

While solar panels are pretty self-explanatory and most people seem to think they’re a good idea (at least on houses’ roofs), heat pumps appear to be more of a divisive topic. Red-faced media outlets are happy to proclaim what a terrible thing they are and how government subsidies to encourage their use are a waste of money. In truth, most people who don’t have a ASHP probably don’t really understand what they do.

In the UK at least, most houses have central heating that’s fired by a boiler heating up and circulating water through radiators or underfloor pipes. At its simplest level, what might happen is that when the controlling thermostat decides the house needs heating, the boiler fires up and heats the water to a high degree – older boilers maybe running at 75C – and then pumps it around to the radiators etc. When that brings the house up to the desired temperature, the boiler will stop. House cools a bit, boiler fires up and the process resumes.

Heat pumps are more granular. The process of warming the water up is like a fridge in reverse, and therefore the amount of electric power that is used will vary depending on the ambient outside temperature and the desired flow temperature of the water. The flow temperature for a heat pump will typically be a lot lower than a gas boiler, and therefore it is commonly held that in order to be effective, you’ll need bigger radiators.

The main difference in practice is that a heat pump doesn’t operate in that on/off mode that a boiler does. Rather than blasting temperature up then shutting down, it wants to be more gradual and run for longer. It can also be more effective at maintaining steady temperature by varying the how much it heats the water. Add in intelligent controls which have both an outside thermometer and even look at the weather forecast, and it can be much smarter than whatever it is replacing (it must be said, there are smart, modern boilers too).

Heat pumps use quite a lot of electricity, so can be expensive to run – but they operate on the principle that they will generate more heat than the electricity user… in other words, by consuming 1kW of electricity, the compressor and pump could generate 3kW or 4kW of heat. This Coefficient of Performance will vary depending on how well insulated the house is, how hot the flow rate is and a variety of other measures.

A case study

Our 30-year old house has never had a natural gas supply. Out front, it had a large pressurised tank storing up to 1,400 litres of LPG (mostly propane), and the gas boiler and cooker had been converted to run on LPG instead of natural gas. In practice, it was “fine”, except that servicing is a bit more complex as engineers need to maintain separate certifications to work on LPG equipment. Also, there was little choice in who to get the LPG from since the supplier owns the tank, and various rules meant it was very difficult to switch from one to another.

Given that we don’t have much choice in the supply, along with the concern that our boiler was surely on its last legs (having had several last rites performed in the past), we decided to switch to the heat pump in early 2026.

LPG is quite energy dense – one litre is equivalent to around 7kWh of power, and pricing wise it works out at about 56p/L with a standing charge of 29p/day. In practice, one litre of LPG would generate about 4kWh of heat energy and during the winter months, we’d be using 7 or 8L of LPG a day. So including standing charges, that would work out about 18p per kWh of heat. A heat pump should be able to generate 1kWh of heat for about 8 or 9p.

It’s a big old unit. This one is a Vaillant aroTHERM plus 10kw affair. It’s tucked around the side of the house and is largely silent when in use. During the hot summer, it was quite nice standing in front of it when it was running (heating water) as the output from the fans is cold air… It’s possible to reverse the direction of heat, so the pump blows out hot air and circulates cold water inside, but we didn’t get that enabled during the height of summer. Maybe I’ll play with it next year.

Anyway, our house has “wet” underfloor heating downstairs, which is ideal for heat pumps to work with. The upstairs radiators did not need to be changed, though the ASHP is probably running at a higher flow temperature than would be most efficient. If you were building a new house, you’d just factor in needing larger radiators from the start.

So far, it’s working well but we’ll find out in the depths of winter how it would compare cost-wise. Graphing our power usage (by kWh) over the last few years, we’ll likely see the 2026 blue bars dramatically overtake previous years since we’ll have the double effect of less solar being generated (so we need to import more from the grid) and the need to use more power to keep the house warm. I’ll do a side/side costing of the switch from LPG in Spring 2027. The electricity in the chart is only power consumed from the grid, not taking into account anything exported – that will feature in the cost comparisons to follow later.

There were some other benefits and additional costs in getting rid of LPG. At the same time, we had to move from a vented, gravity fed water system to a pressurised closed one. That means we have strong water pressure at every tap, though also had to replace shower pumps. The hot water cylinder moved from an airing cupboard into the loft, meaning we have more storage space too, and we don’t need loft space for big water tanks any more. The old LPG cooker went on eBay and got replaced with (an expensive!) shiny new electric one.

And I was glad to see the old LPG tank be craned away. Still not sure what to do with the space it’s freed up… maybe space to store another car…?

Costs

We used a company called Heat Geek, who are an umbrella for the supply and installation of the heat pumps. They have some bespoke software that’s used to scan the thermal efficiency of the house, and supposedly come up with a guaranteed level of performance based on their modelling.

The total cost for everything to do with heat pump install, worked out at a little over £10,000 (including the government Boiler Upgrade Scheme grant of £7,500; as it happens, a few weeks after our system went in, we could have got another £1,500 grant since we were moving off LPG… oh well). If we’d had to replace the old LPG boiler, it would probably have been in the £3-4,000 region.

If you’re in the UK and thinking of making this switch, you’d be better off doing it before the next General Election. If there’s a change of government, the BUS is likely to be sacrificed as being too woke and net zero.

The new cooker was £1,000s and there were some associated costs in getting rid of the LPG tank, so we’re probably in for £15k spend in total. All in, I’m happy. The systems are working well and are much more tech enabled than anything we’ve had before, allowing me to fritter away time tweaking Home Assistant…

I’ll report back in the spring to share how the winter went, in case anyone is interested 😊

New Yorker cartoon by Alex Gregory, 25 May 2015

The media on which people consume music and video has evolved through clear phases in its history. The wax cylinder gave way to spinning vinyl, a variety of tape formats competed and were ultimately superseded by shiny plastic discs read by lasers. Ultimately, all have been deemed irrelevant by streaming and digital storage but even allowing for the “format wars” that have run for sometimes decades, this is one of a few examples where users are choosing technology which is regressing.

Main cartoon image source: New Yorker, May 25 2015

The music media wars

By the 1960s, the dominant form of music that people consumed was vinyl, particularly 7” singles. Radio of the time wasn’t very good quality – but that didn’t stop the Rock & Roll  revolution powered by scratchy AM transmissions from the likes of WINS 1010 in New York or Radio Luxembourg broadcasting across Free Europe.

The kids would listen to the radio then rush to buy the latest records to play on their Dansette, in glorious mono. Some car manufacturers even fitted record players under the dash (amazingly, even until the mid-1970s…)

Source: When Cars Had Turntables: Photos of the Era When Vinyl Hit the Road, 1950s-1960s – Rare Historical Photos

The 8-track cassette cartridge had a brief moment in the sun, especially in-car, and had some theoretical advantages over other formats (like support for quadrophonic sound), at least at first.

The Compact Cassette arrived in the mid-60s, allowing people to record stuff off the radio or from their own records, perhaps making compilations or “mix tapes” for themselves or to share. Cassettes were smaller and much more usable in cars – and as the composition of the tapes themselves advanced, they trumpeted improved sound quality. Technology like Dolby noise reduction improved things too.

It was the arrival of the Sony Walkman and its ilk in the early 1980s really cemented the cassette tape as the primary format, for pre-recorded music as well as DIY.

Source: Google Gemini

So, the pattern is set – a format or technology that has evolved as much as it can ultimately gets overtaken by another one, even if the quality isn’t initially as good. The same thing happens all over – see the 1st chapter of The Innovators Dilemma, which shows 5.25″ hard disks being replaced by initially less good 3.5″ ones, upending the storage industry of the time. When a new technology gets at least as good as a previous one, and becomes more convenient and/or less expensive, it seems like a one-way journey to its adoption.

Audiophiles in the 1980s would point out that cassettes never sound as HiFi as their vinyl, even before the tape gets accidentally scrunched, stretched or dusty. Quality doesn’t always matter as much as convenience, and the tape was cheaper and more easier for most people.

You can see where this is going. The Compact Disc arrived in early 1980s and despite being initially more expensive than both vinyl and cassette, quickly established dominance. Objectively, it does sound better, it doesn’t need rewinding and is supposedly dust and scratch resistant. Oh, and you can random access a CD to skip tracks without faffing around with ffwding or dropping a needle.

By the mid-1990s, vinyl had flatlined to irrelevance, CDs were still booming and tapes were starting to lose their place, at least as a medium for recorded music. Arguably, this was the best time for mix-tape creators as it was easy to record tapes from borrowed CD sources that would sound as good as pre-recorded cassettes might, so sales of blank cassettes were still peaking.

Fast forward another 20 years though, and the format had changed again – and we’re not even looking at streaming services. When MP3 appeared in the mid 1990s, it was at the forefront of digital music. Storage devices were still low-capacity and expensive, especially portable ones – at the time a 1GB hard drive would cost well over $1,000. Even by 2000, the relatively new SD card was $200 for 64MB … enough to store maybe a couple of dozen MP3s?

The dramatic fall in costs and rise in capacity of mobile storage, combined with easy sharing (via Napster or just giving copies to friends) allowed digital downloads to massively overtake and displace CDs. Portable MP3 players like the iPod killed the cassette almost entirely.

Digital formats

There was a time when the format that digital music was stored in echoed previous commercial battles within the same basic device type – VHS vs Betamax for video cassettes (or even V2000 if you can remember that), Laserdisc vs DVD, Blu-ray vs HD DVD etc.

MP3 was effective at reducing the size of stored music, but it is “lossy” – arguably, it throws away the data that you don’t really need, in favour of being more convenient (i.e. You can fit 20 or 30 MP3s on a media card, rather than 1 or 2 uncompressed files).

Microsoft tried for a while to cement its own standard, Windows Media Audio (WMA), as an alternative. Using a better mechanism to do the compression (and have the option to be lossy or lossless depending on your preference), WMA could create smaller files that sounded better than MP3, at least if the bitrate of the conversion was low. Microsoft saw WMA as a licensing opportunity, to the point where lots of hardware devices that played MP3 could also play WMA files – CD players reading files off a CD-ROM for example, useful in-car when you could fit 100+ files on a CD instead of only 70-odd mins of music on an audio disc.

There was also a misguided attempt to shoehorn Digital Rights Management into WMA, meaning that if you used Windows Media Player to “rip” a CD on your PC, it would create WMA files that had rights protection applied. The intent was to try to prevent illicit copying, and Microsoft had a certification and licensing program for 3rd party devices called PlaysForSure, meaning you could legitimately copy protected content (such as tracks bought from its MSN Music store) to supported 3rd party devices.

Before Apple’s iPod came along and swept everything before it, there were a variety of portable media players – Creative Labs pioneered the HDD-based Nomad in 2000 and the range evolved in competition to iPod for a few years. One example is this Creative Zen Micro music player c2005, which had a 4GB hard disk spinning away inside.

Somewhat ironically, when Microsoft launched its own late-to-market music player, the ill-fated Zune in 2006, it was not compatible with its own PlaysForSure standard. So, if you’d already bought and downloaded protected content from MSN Music, you couldn’t load it onto your new Zune without buying it again from the Zune Marketplace.

The lesson? Don’t trust “owning” anything that is dependent on a company allowing you to continue to play it… MSN Music and Zune are now all gone and if you come across media that was protected by them in the past, you’re unlikely to be able to play it since the services that validated the license are not there anymore. Microsoft did publish a tool that might help you un-DRM your own media though… good luck with that.

Amazingly, MP3 players are staging a comeback, of sorts. If you have £130 doing nothing, you can also buy a rechargeable Bluetooth capable cassette player.

The Renaissance

In an unlikely twist, vinyl record sales started to take off again in the mid-2010s. Digital streaming is, by a long way, the biggest player – in the US it now accounts for 82% of revenue vs 12% for the total of physical media.

Nowadays, vinyl accounts for an ever-increasing slice of the pie. Although unit volume in the US overtook CDs in 2022, in the UK, CDs still outsell vinyl (just): in the first half of 2026, 4.2M CDs were sold, compared to 3.8M vinyl LPs. But the value is telling – with the average price of an LP being more than double that of a CD, that’s where the money is. In fact, in the US, vinyl accounts for 75% of physical media revenue, so 9% of total music revenue.

The trend in the UK is lagging a little – vinyl sales by volume is still growing by 15%+ YoY and CDs are shrinking by ~5%. Even though some artists are releasing new music on vinyl, CD and even cassette, most new stuff goes straight to streaming platforms.

And most listening does too…

Do you remember the time before everyone had a mobile phone? Even what it was like to not have 24×7 internet access? It is nearing on 20 years since the smartphone revolution as we know it started, with the release of the iPhone in mid-2007.

History is littered with technologies which quickly faded, after everyone saw what followed them. In the mobile phone world, once people realised what they could do with a big, colourful screen, faster-and-faster mobile internet coverage and ubiquitous Wi-Fi, everything changed.

Not for the first time, Microsoft was a real leader in the field but managed to fumble the ball, while Apple and later Google harnessed fast-developing technology to give people an experience that was like magic compared to what they were used to.

The origin story of Windows Mobile

In the mid/late 1990s, Microsoft built a modular operating system called Windows CE (the CE bit doesn’t stand for anything, apparently – it’s sometimes thought it was “Compact Edition” but that’s not true). It could target a variety of different hardware types, since Intel x86 wasn’t really viable on a pocket-sized, battery-powered device, and CE could be heavily customized by the OEM using it. Essentially, a device maker could roll their own operating system using just the bits they needed for their specific hardware; it might have a screen, it might have a network, or it could be something totally custom.

Microsoft itself came up with some reference categories for mobile devices using CE underneath, the first of which was the Handheld PC.

HP Jornada 720 Handheld PC – one of the later and more capable “HPCs”

The thinking was that people could have a smaller, mobile device running something that looked a bit like Windows (albeit smaller in scale). It might use a stylus to control it, and it could be licensed in the same way that PCs were – i.e. the OEMs would pay Microsoft $10 or whatever for every device they sold. Microsoft would bundle a load of built-in apps like shrunken versions of Office, and 3rd parties might build their own apps to run on it too.

The initial batch of devices were seen as companions to the PC and communicated over a relatively slow cable or even using Infra-Red (remember IrDA?). Bear in mind that in the late 1990s, Wi-Fi as we know it didn’t exist and internet access from any mobile device would have been dial-up modem speed. GPRS data comms didn’t arrive until 2001, and even that was in the realm of 50-60 kilobits/second.

Competition for these companions were mostly things like Psion organisers or PalmPilots (if you don’t mind a bit of NSFW humour, check out what Sweary Brian Blessed had to say about them). It was a market that was growing amongst an enthusiastic user base, mostly tech-savvy and business-oriented and doing the kind of thing you might have had a Filofax for previously.

Other form factors

Handheld PCs evolved and were joined by other Windows CE form factors, like the Pocket PC, with Compaq’s iPaq being the most successful, arriving in early 2000.

The iPaq 3630 was one of the first consumer devices with a Lithium-ion battery, making it surprisingly light and powerful. If you wanted to expand usage, you could buy “jackets” that incorporated additional battery capacity and offered CF memory cards or PCMCIA slots to plug in Wi-Fi cards, modems and the like.

May 2001 – Some Microsoft spiv showing Tony & Cherie Blair around an iPaq with PCMCIA Wi-Fi card in a jacket, half the size and weight of a house brick

The Developer promise

Microsoft brass at the time could see the revenue opportunity of licensing Windows CE to handheld device makers, and the highly modular nature of Windows CE could be turned to other things like “headless” operations where it functioned just as an embedded control system. The Sega Dreamcast console famously used Windows CE and Microsoft had an effort (later called Windows Automotive) where they tried to persuade car companies to fit their software into the car.

If you’re interested, delve more into the history of this dead-end on The time Microsoft made a car stereo – The Auto PC Story, presented by ex-MSFTer, Andy Haslam.

Anyway, the architecture of Windows CE was such that it looked familiar-ish to developers who were used to writing software for Windows. Even though the underlying hardware was very different, you could build CE-targeted apps using Visual Studio, test them on the PC and then deploy them to your connected device.

At the time, some 3rd party apps were available for Windows CE devices – some licensed and some free. Generally, they were deployed by downloading an installer to the PC, then docking the device so the app could be side-loaded. At the time, a perfectly reasonable way of doing things. A UK company called Visual IT built a widely-used Tube app, to allow offline planning on the London underground. Everyone had it but now there isn’t even a screenshot of it online.

When Apple launched the iPhone in 2007 and then the App Store a year later, it revolutionized how you could get software quickly and at scale onto mobile devices. After that, Microsoft lost any advantage they might have had with developers already knowing their tools and some of the technologies shared between Windows and Windows CE. Professional developers flocked to the millions of iPhone users within a year or two of its initial release, even if the actual development environment was less than ideal.

The Windows Mobile Family grows

Windows Mobile was the collective name given to the variety of mobile devices Microsoft was working on (though during their gestation, they had many different branding eras). It’s a noteworthy distinction – Windows CE was the operating system that could be applied to all sorts of things, and Handheld PC, Pocket PC etc were software layers that sat on top, and defined a category of device that would be manufactured by a host of OEMs.

Archive.org’s snapshot of the Microsoft Mobile website, April 2003

In early 2002, the first of a whole new category of product was also launched – the Windows® Powered Smartphone (branding simplicity was never a Microsoft strength). At that point, mobile phones were either “candybar” (i.e. a single shape with a screen and some buttons) or “flip” (like the Motorola Razr). Some strayed a bit –  like the Nokia “banana” phone with a slide-out cover over the keys (fun fact – when Neo answers the phone in the Matrix, it’s spring-loaded where the regular phone wasn’t… thereafter every 8110 user tried to flick it open because it looked so cool).

When the original Smartphone was launched, it featured a small colour screen and a keypad with a joystick/cursor pad for moving around the various menus. It was functional, and gave easy access to contacts, email and calendar, all the time using GPRS (aka 2.5G) to keep up to date.

Getting into the phone business was a complex affair – for premium handsets, the phone would invariably be bundled with a contract, meaning the device was really owned by the network at least until the contracted period was up. As a result, the networks expected to have complete control over how it worked, when it was updated and even what features should be enabled or not. Given that they were supplying and supporting the devices, that seemed fair enough at the time, though in hindsight one of the most revolutionary things about the iPhone (5 years later) was how it empowered the device maker over the network operator.

The first Smartphone launched worldwide was the Orange SPV, in London, in late 2002.

It really was an exciting time since there was nothing really comparable on the market, unless you strayed into Blackberry for being a portable email machine, or the various Nokia type phones (“feature phones”) for being really good at phoning & texting and having battery life measured by the week.

Motorola launched a flipphone version of the Windows Mobile Smartphone …

… and a variety of handsets continued to evolve (the Orange SPV C500, aka HTC Typhoon, becoming standard issue to everyone in Microsoft UK)…

Meanwhile, in the touchscreen, larger sized device world, there were also developments.

Microsoft built phone capability to run on the Pocket PC, essentially extending what had originally been an offline device to have a built-in radio, and the phone & messaging software to go with it. Compared to the Smartphone it was a much less good phone, and the battery life was not good; but then it had a nice big screen and could do a lot more…

Launched originally as the O2 XDA (and the catalyst for the device hacking community which still exists, XDA Developers), it heralded a load of larger and more capable devices which could be used for more than just the basics.

Both Smartphone and Pocket PC Phone evolved further, sprouting devices with slide-out keyboards (to compete with Blackberry).

HTC Vox

HTC Herald

Steve Ballmer famously mocked the original iPhone because it was expensive and didn’t have a keyboard so businesses wouldn’t want it. Right.

HTC was a very capable OEM who was happy to churn out devices that would be picked up and customized by the various networks; the Orange SPV was really an HTC Canary; the O2 XDA an HTC Wallaby and so on. The ultimate Pocket PC was arguably the HTC Universal, a clamshell device with twist & flip capability that had a forward-facing camera and was the first such device available supporting 3G/UMTS.

The networks were still too slow, the hardware wasn’t really capable enough, it was ferociously expensive, heavy and the battery life was terrible. But apart from that, it was fine.

Windows Mobile stumbled on until 2009, with its final release trying to catch up with the iPhone. Android was only getting started, with HTC making the very first device with a similar set of “throw everything at it” kind of design that was a hallmark of Windows Mobile.

Microsoft went on to throw Windows CE and the whole Windows Mobile ecosystem away, starting again with a whole new OS, hardware design and app development ecosystem. Windows Phone 7 was released to critical acclaim in late 2010, but it was too late. Android was getting a head of steam and Apple had already launched the iPhone 4. What might have been?

Anyone who works for, or closely with, Microsoft will be aware that their fiscal year starts on 1st July. The first couple of months of the new FY are typically consumed with reorganization, internal or channel sales kickoff type events (like MCAPS Start for Partners) then going on summer holiday. Basically, the year starts for lots of people in September.

If you are a partner company to Microsoft – or any other major company for that matter – it pays to be very aware of where they are in their FY cycle. John Rudow, who has artfully managed to sell more-or-less the same kind of training materials to different roles within Microsoft for 20 years, recommends figuring out what the “mutual operating period” is – in other words, when is each company thinking about plans for the next year vs heads-down in execution?

Since Microsoft’s Q4 is April-June, if your FY is April – March, don’t be surprised if the first half of your year is pretty much gone before you can hit the sweet spot of working alongside Microsoft. If you’re trying to get Microsoft’s attention for what you do, then the end of June is a bad time, as everyone in their sales organization is running around trying to get stuff over the line.

FY26 is a wrap

The results for the FY ending 30 June 2026 have just been announced; if you’re a shareholder then you’re probably smiling. The pattern often seems to be that the company announces record results (at least using the measures it chooses to report) and the share price falls, because the upside is already supposedly priced in and there’s often a downside that can be dwelt on. Lots of capital expenditure on AI datacenters but not enough uptake of Copilot? Tsk tsk. Azure growth slowing from stratospheric to merely incredible? Oh dear!

This time round, the FY26 Q4 results were pretty stellar and beat analyst expectations in a number of areas –

• Revenue was $90Bn, up 18% from the previous year; analysts were expecting around $87Bn
• Adjusted Earnings per share (EPS) was $4.74, up 23% YoY; analysts expected around $4.30.
• “Intelligent Cloud” revenue grew to $39.3Bn, up 32% vs expectations around $38Bn.
• “Productivity & Business Processes” revenue $37.8B, up 14% YoY, against expectation of around $37B
• “More Personal Computing” actually fell 4% at $12.9B, but remained slightly ahead of expectations.

There were a few standout metrics in both Q4 (which everyone knows is the busy time for Microsoft as customer deals tend to slide into June) and for FY26 as a whole, like Azure selling over $100Bn in the year, growing 43% YoY for Q4. That is insane levels of growth on an already large and mature business.

Here is the detail of Microsoft’s Q4 release; fine if you like eating dry crackers for your mid-morning snack. If you prefer more-digestible chocolatey biscuits, check out Jack Rowbotham’s excellent post which also includes AppEconomyInsights’ great summary diagram to show where the money comes from and where it goes.

The 3 year picture

For the full FY26 fiscal year, the Big Dog is actually Productivity & Business Processes, which includes most of M365, Dynamics, and LinkedIn). Over the last 3 years, it’s grown $33Bn in revenue and an incredible $24Bn in income.

Intelligent Cloud managed to add $51Bn in revenue over the same time, though income has “only” grown by $19Bn. It shows you where the real margin is.

Even if the news wasn’t quite as bad as analysts were expecting, More Personal Computing has been pretty flat over the 3 years; that includes Xbox consoles and games/services, Surface devices as well as the revenue from all those lovely adverts you might see in Widgets, Bing, MSN and the likes. Oh, and Windows. Let’s not forget about that – the cornerstone of the whole company, without which there would be no Microsoft today.

They don’t break out how much revenue or profit is delivered by Windows client (Server sits in Intelligent Cloud) – but it’s going to be pretty high margin stuff. Analysts reckon that Windows delivers about 5% of all revenue, but that profit for this year would be in the $10Bn region. Breaking out Windows from More Personal Computing makes things interesting.

So, Microsoft still makes a load of cash from Windows, even though it’s not exactly been strategically front and centre in recent years. Office and all its hangers on in M365 is insanely profitable and Azure, with its $100Bn+ annual revenues, is a cash printing machine.

Share price & market value

In May 2025, Tip of the Week #76 commented that Microsoft had the largest market capitalization of any company – a total value of $3.33 trillion, marginally ahead of Apple at $3.2T and NVIDIA “only” $10B behind in third. Amazon was a seemingly distant 4th at $2.24T (so around $1T less that Apple), while Alphabet/Google was 5th at around $2T.

Exactly a year later from ToW #76, Google’s share price was nearly 2.5x what it had been previously, meaning their market cap was $4.8T. By any measure, that kind of growth is remarkable. NVIDIA was sitting at $5.5T, and has grown by nearly 10x over the last 5 years.

Source – Stock Analysis Market Capitalization comparison

And since Market Cap is the number of outstanding shares multiplied by share price, the comparisons of share price show a similar picture. As of mid-afternoon UK time, 4 August:

Source: Stock Analysis stock price comparison

Traditional valuation of stock prices would be based on things like P/E ratio, the forward outlook (like are they seeing great opportunity or declining sales, are they spending more than they earn etc) but actual logic and share prices seem to have been divergent for a long time. Surely there’s a bubble here; Google/Alphabet’s share price on 15 May 2026 was $393.32. On 15 May 2023 it was $116.96.

Between FY23 [since Alphabet’s fiscal year is the calendar, useful info you need to have a Mutual Operating Period?] and FY25, they grew revenue by 31% and operating income by 53% – remarkable numbers. But should they justify the company being nearly 3x more valuable today ($4.55T) than they were 3 years ago ($1.61T)? Even if there’s a crash and all of these multi-trillion companies shed 2/3 of their value, will they still be over-valued compared to others who don’t inhabit the same spheres?

Look at car makers for comparison – here’s the market cap for Ford, General Motors, BMW, Mercedes and Volkswagen over the last 5 years:

Stock Analysist Market Capitalization over 5 years

But add Tesla to the mix:

Those 5 traditional car companies have a combined market cap of around $250Bn, while Tesla’s is 5 times that. Yet the traditional manufacturers’ combined revenue is 10 times that of Tesla and operating income (EBIT) is estimated to be slightly better for the traditional automakers – it’s about 11x that of Tesla.

Now, traditional car companies are losing money on lots of new endeavours so even if their top line looks healthy, the future is a lot less promising.

But still; how can the market see be such a big gap in valuations? Answers on a postcard, please…