Software giant Microsoft (MSFT -1.84%) is transitioning into a devices and services company, but the "device" part of the equation hasn't been going so well. While the company's proposed acquisition of Nokia's phone operations is a smart move, Microsoft's tablet strategy has thus far been lacking. In July, the company took a $900 million inventory write-off associated with its Surface RT tablet, a product which runs the ARM-based Windows RT. Total Surface revenue through the second quarter, which includes both the Surface Pro and Surface RT, was actually smaller than the size of this write-off.

After the terrible performance of the first generation of Surface devices, Microsoft has announced new versions of both the Surface RT, now called the Surface 2, and Surface Pro. While Microsoft has made some much-needed improvements that should help sales going forward, there are still some serious flaws in the company's tablet strategy.

The good: Surface Pro 2
The Surface Pro 2 is meant to be a laptop replacement, appealing to businesses and power users. It's not a mainstream tablet, starting at a hefty $899. Instead, the Surface Pro 2 is aimed at those who do more than browse the Internet and play games. It competes more with laptops and ultrabooks than it does with the iPad.

The original Surface Pro, while more successful than Surface RT, had its share of issues. The biggest problem was battery life. Under a full load, the original surface maxed out at about 2.5 hours, somewhat negating the mobility of the device. The Surface Pro 2, by using a Haswell processor from Intel (INTC -1.79%), has increased battery life by as much as 75%.

The other issue was Windows 8 itself. While the Surface Pro ran the full version of Windows 8, the operating system had some issues that drove people away. Windows 8.1, an update set to be released in October, looks to fix many of these issues. The start button will be brought back, although it will simply lead to the start screen, and users will be able to boot directly to the desktop. The Surface Pro 2 will ship with Windows 8.1, but whether or not the changes will be enough to win people over remains to be seen.

New, slimmer covers are launching with the tablets as well, reducing the bulk while adding backlighting to the built-in keyboards. The new Power Cover comes with both a keyboard and a built-in battery that increases the battery life of the Surface Pro 2 by about 50%. With this, the battery-life problem of the original device is effectively solved, although it does require separately purchasing the Power Cover.

The Surface Pro 2 is not a mass-market device. It doesn't really compete with the iPad or tablets running Google's (GOOGL 0.35%) Android OS, instead looking to replace the laptops of power users. The new Surface tablets include Microsoft Office, and with Windows already dominating enterprise PCs, the Surface Pro 2 may see significant enterprise sales, especially now that battery life has been improved. I think that Microsoft will sell enough of these tablets for it to make sense to continue making them, but it definitely doesn't have the mass-market appeal of less expensive options.

The bad: Surface 2
The Surface 2 is the second iteration of the Surface RT, running Windows RT and powered by a NVIDIA Tegra processor. Windows RT, also set to receive and update along with Windows 8, features only the modern tile-based interface and eschews the desktop entirely. Normal Windows applications can't be run on RT, with the Windows Store the only source of apps.

The new Surface 2 includes 25% better battery life and an upgraded screen. It sells for $449. The problem is that the Surface 2 doesn't fix the fundamental flaw: Windows RT. The Surface 2 competes directly with Android tablets and the iPad, but with an inferior app store and no support for legacy Windows applications, along with being more expensive than many other options. There doesn't seem to be a reason to choose the Surface 2 over the competition. Google's Nexus 7 costs $229, for example, and the cheapest full-size iPad from Apple is only $499.

Microsoft should have killed off Windows RT after the disastrous inventory write-off, or at least not continued to make RT-based tablets that no one wants. Investors can hope the company has made less of them this time around, recognizing that demand isn't very high.

Mobile Wintel
While the Surface Pro 2 is aimed at power users, mainstream Windows 8 tablets will need to be cheap in order to compete with the Android offerings. At the same time, Intel needs Windows 8 tablets to gain popularity in order to push its mobile processors, although Intel chips can also run Android. Intel has stated that its goal is to enable tablets in the $100 range, and its Atom processors may eventually make that a reality. The recently released Bay Trail processors, the newest Atom iteration, are impressive. The Z3770 was tested at Anandtech and shown to be faster running Android than any ARM-based processor on the market, all while using roughly the same amount of power. This should lead to some design wins in the high end for Intel.

This also means that Windows-based tablets with Intel chips will be faster than competing ARM-based Android tablets. With high-end Android tablets priced around the $500 mark, a more powerful Windows tablet at a lower price could conceivably sell very well. ARM processors have always been about low power, but now that Intel has achieved acceptable power efficiency, Intel's offerings should trounce ARM-based chips in terms of power-per-watt going forward.

The bottom line
Microsoft continues to make the same mistake with the Surface 2, a tablet which no one really wants. The Surface Pro 2 is a powerful laptop replacement, and the upgrades should compel power users to take a second look. Going forward, Intel's low-power, high-performance mobile chips are the key to getting Windows tablets at prices that are competitive with Android tablets, and the first step in that process is currently taking place. I'm optimistic about Microsoft's future in tablets, but Windows RT will likely not be part of that future.