I've been fielding a lot of questions lately about how the Internet should be funded and managed. It's a great question, because it's not clear that the current model is sustainable.

Quite simply, it may cost more to operate the Internet backbones than the carriers earn - and running at a negative profit margin, as we all know, eventually results in bankruptcy.

Service providers are reluctant to break down revenue by product line, so it's challenging to tease out real numbers regarding the actual cost (and margins) of operating Internet backbones. (Anyone who's prepared to deliver such figures - confidentially or otherwise - please contact me.)

That said, it's pretty straightforward to back into some provocative statistics: Of the typical company's telecom spend, roughly half is on what I'll call "core voice and data services." The rest goes toward remote access, mobile connectivity, hosting, managed services and other offerings. But the percentage of enterprise spend that's purely Internet is typically in the single digits.

Moreover, although the consumer Internet access market is roughly $25 billion and growing in the high single digits, according to research firm In-Stat, that money goes to Internet access providers - the DSL and cable companies - not the backbone providers. And though access providers pay backbone providers via peering agreements, peering doesn't always cover the true cost of providing backbone services.

The bottom line is that money is pouring into Internet access services, but Internet backbone services are being subsidised by business voice and data. And here's the gotcha: Voice margins - and to a lesser extent, data margins - are collapsing, meaning that carriers will at some point soon find themselves unable to continue subsidising the unprofitable backbone business with the no-longer-so-profitable voice business.

There are typically two responses to this scenario. Some of my telecom colleagues argue for nationalisation of the Internet infrastructure, on the theory that a universally reliable Internet is as valuable as, say, roads and public transportation systems.

Others argue nationalisation is unnecessary and harmful, and that the free market will adapt to enable providers to offer low-cost, high-quality backbone services.

So what's the answer: Nationalisation or blanket reliance on the free market? My vote is none of the above.

Try this thought experiment: Let's say the Internet went dark tomorrow. Who'd stand to lose the most? How about those companies whose entire business models rely on a functioning, widely available Internet: Google, Amazon, eBay and their ilk. These three alone have annual revenue of $13 billion.

It's just a guess, but I'd be willing to bet that these firms - and companies like them - would welcome an initiative to put in place a framework that would allow carriers to set and receive fair-market payment for backbone services, in exchange for a commitment to deliver highly reliable end-to-end connectivity for every consumer who's willing to pay for it.

In the past, I've discussed the proposal for an International Association of Networking Service Providers that would create such a framework. My question then was: Who cares? I'm guessing that Google, Amazon and eBay do.

Johnson is president and chief research officer at Nemertes Research, an independent technology research firm.