For the past two days I’ve been asking myself why Siri is limited for iPhone 4S. It seems to be purely arbitrary. It’s not because the preconditions are not available: The iPad 2 has everything what the new iPhone consists of. A possible explanation: Apple wants to start the necessary server farm for Siri with a controlled market launch of its devices.

My assumption is a different one: It’s the costs.

Like I’ve illustrated a few days: The heart of Siri is definitely not the (surprisingly good working) speech recognition. The highlight is the AI machine underneath. Each question to Siri triggers a background communication between the iPhone 4S and the central cloud of Apple.

But what does a Siri dialogue cost? Nobody knows but Apple. Nevertheless, I think it is fair to assume that a Siri dialogue will have the estimated costs of a Google request. There are fairly reasonable ratings which classify these costs at about 1 cent. Countercheck: Companies which want to use Google as onsite service are paying 2 cents.

So I think the costs of 1 cent for a Siri request are rather the lower bar (especially if you consider that Google has a 10 year experience to optimize its central data centre regarding its costs).

And now it gets exciting. How many Siri dialogues does a regular iPhone user have? That’s something Apple would like to know as well. A model calculation: 10 daily dialogues make 3,650 dialogues per year. Multiplied by 1 cent it sums up to $ 36 per year. No problem for Apple with a a margin of 70 % for the current iPhone and a period of use of three years.

And this a problem for Google and its Android eco system. Google is refinancing Android with its mobile search revenues. These run up to approx. $ 10 per year and device. Further profits via licenses or hardware do not (yet) exist. In the end that is a whole lot of money counting the pure number of devices only – but it’s not enough to run a profitable cloud-based assistance service like Siri. If Google would start a Siri clone, it would mean a loss of $ 20 per device and year. At the same time they would loose the text-based adword revenues. Again, it’s the quantity that counts – this time with reversed signs.

If consumers will love Siri it will – next to an extremely strong lock-in – also cause a big problem for Google’s mobile business. A cloud-based assistance service is nowadays not fundable without gigantic hardware margins. And which provider owns these other than Apple?

Ironically, a too big success cannot be afforded by Apple either. As long as the server costs will not go below 1 cent/dialogue, Apple would get into trouble. If e.g. 100 instead of 10 daily dialogues will take place between an iPhone and a Siri machine, the nice margins would be gone in an instant. In this case Apple would have to start the development of an independent business model for its Siri usage right away.