Robert Pogson

I think it is clear that 2009 was the year of GNU/Linux on the desktop:

• Most OEM’s supply units
• Doing very well on netbooks
• That will continue with ARM
• Many plans are being made to migrate from XP to something and in many cases it is GNU/Linux
• Even the most stodgy are giving GNU/Linux a look
• No one can claim it is not ready
• Virtualization continues the trend with thin clients and more
• Cloud computing works with GNU/Linux

That’s water under the bridge. What can we look for in 2010? Here are my predicitions:

1. ARM will cause large price moves for netbooks and other manner of PCs. There is nothing to stop this as price is a major pleaser for consumers. If established OEMs boycott ARM they will lose to dozens of smaller operators. With such narrow margins in hardware, no one can afford to lose anything to anybody. Acer has overtaken Dell. Will Dell be willing to stay at Second Fiddle? Acer is not afraid of the low-end market.
2. Oracle+Sun will complete. What has this to do with the desktop? Servers and thin clients make the best desktops for most (but not all) tasks. Sun makes great servers and will continue to do that. SUN VirtualBox will work for many of the fancy virtual desktop scenarios now looking good. Thin clients will run this show and any OS will do. UNIX-like OS with shared memory for host+guest are optimal in performance. GNU/Linux on the terminal is a natural because there is no per-seat charge. Business at the large end see this now. The middle will see it in 2010. The competition will try to do the same.
3. The Linux KVM will mature making sure GNU/Linux fits in all the virtualization tricks in 2010.
4. AMD will awaken and realize it must innovate to take share from Intel. They did that when they went to 64bits. That was the right move towards share in servers and high-end desktops and HPC. They must innovate and get back into the 32bit game for thin-clients and small PCs. For the same reasons that 64bit makes sense, they should consider 24bit for the small format machines. They don’t need a gigabit of RAM to do the job sometimes. Maybe 24bit is extreme but less than 32 is not. This is a stretch as AMD shows no interest in small in 2009. They need to cut power/bits/die size to compete at the low end in 2010. Smaller dies are quicker to design. They could be in production by June of decent competition for Atom/Arm. In 2010, the low end will rule: netbooks, smart-whatevers, notebooks, thin clients all will want less-is-more hardware costing less and using less power. At 22nm, AMD could take serious share and grow a market in which Intel only dabbles. Intel has to protect the Wintel monopoly in which it has invested. AMD has nothing to lose from Wintel.
5. VIA has had a hard hit this year as Atom pushed them out of netbooks with more power. VIA has to innovate by getting to higher resolution dies to increase productivity while cutting power and increasing speed a bit. They are a little too slow for netbooks at 400 MHz and too expensive for thin clients for $100 just for the motherboard. I predict they will find ways to cut prices while increasing volume in 2010. VIA can compete on price if GNU/Linux rides on top. Expect more partnerships with VIA working to distibute GNU/Linux on their kit.
6. Intel will have to go to higher resolution to compete against AMD etc. on the low-end. Nothing prevents them from producing ARM chips for the low-end. They, like everyone else will need to use GNU/Linux to get the performance buyers want from low-end devices. They are predicting 32nm. In 2010 they will predict 22 nm and soon. ARM is very competitive now at 65nm.
7. 2010 will be the year of the thin client. IBM, RedHat, Novell, HP, Dell and many others are positioned to supply thin clients in volume in 2010. Rate of increase in production has been high for years but now they are too useful to ignore. Large business will deploy many millions in 2010 just to get off the Wintel hardware treadmill. If they feel pressure to move from XP to new hardware, it makes sense to use thin clients. Then the OS on the thin client does not matter and GNU/Linux will be a cheaper solution.Thin client production will pass Mac production in 2010.
8. SSD will be competitive on price performance with magneitic discs in 2010. They are superior in performance but still too expensive in 2009.
9. There will be an anti-trust law suit filed against M$ in 2010 over GNU/Linux. Enough businesses selling GNU/Linux are losing enough money because M$ is campaigning against GNU/Linux that this is feasible. For example, when M$ provides training materials for retailers’ staff putting down GNU/Linux and denying a market for OEMs selling GNU/Linux, this is grounds for an anti-trust case. Same goes for prohibiting  benchmarking, selling same hardware with GNU/Linux and M$’s stuff, and revealing the cost of the OS.
10. Several large US governments will convert to GNU/Linux thin clients in 2010. It just makes sense. The USA has supported M$ further than there is any duty or economic interest. Arguments of protecting US jobs ring hollow in the face of cointinually rising costs for malware, downtime, patching, etc. Taxpayers should have a say, too. They are tired of tax dollars flowing down the licensing funnel to M$. Munich may never pay another volume licence to M$. Why should NYC or LA?
11. M$ will not show a rebound in the client division as long as it throws money at netbooks to keep out GNU/Linux. Ballmer will give up that battle with no long-term benefit to the company in 2010.
12. The SCOTUS will throw out software patents in 2010 in support of the US constitution which provided copyright for published works and patents for inventions. Software is a published work, not an invention. It is a logical stream of instructions producible by anyone skilled in the art given the desired end-result.