The telecommunication future may still look Orange, or Vodafone or even DoCoMo, but it surely doesn't look as bright as it did 12 months ago.

Consider this: even as the network operators themselves are consolidating into bigger and fewer players, the head of DoCoMo's wireless internet service, i-mode, worries that mobile networks of the future may not provide the revenue growth that telcos are betting on. He should know. With nearly 16m subscribers signed up since its launch in February 1999, Japan's i-mode has been a runaway success.

I-mode is adding 1.4m subscribers every month and is predicted to reach 20m subscribers by March. Of DoCoMo's 33m wireless customers, nearly half use i-mode. Its success stands in stark contrast with the disappointing rollout of similar services in Europe and the US called Wireless Application Protocol, or Wap.

In America, a paltry 2% of AT&T's 15m wireless users connect to the internet using their mobile phones. Sprint's PCS unit has done better with some 9% of its 8m wireless customers attempting to dial up the net while on the move. In the UK BT Cellnet trumpeted earlier in the year that it had cornered the market for Wap-enabled handsets for the UK, only to concede in October that no more than a third of its new mobile customers were buying Wap phones.

Chalk it up to Wap-lash. The operators and handset makers hyped Wap as the new Coca-Cola. It was going to take the market by storm. Unfortunately, the circuit-switched networks were too slow for anything other than simple text-based messages.

What the European and US operators have been telling us is that higher-speed networks - called GPRS or 2.5G and UMTS or 3G - will right all of Wap's wrongs. The unnerving thing is that i-mode's chief Keiichi Enoki disagrees.

He thinks the revenue models of 2G and 3G are going to be much the same because even with faster speeds, sustained bursts of content will be hampered by lack of spectrum capacity.

While the bandwidth will be greater, the number of users and the size of the applications will be greater too, effectively leaving users with the same bandwidth as 2G. In this scenario, Enoki says that 3G devices will be best suited for sending song snippets and 10-second video clips to users to whet their appetite. Then the same users will wait until they are at their fixed-line, broadband-connected PC or interactive TV before downloading whole songs or movies.

In this case, who gets the revenue? Not the wireless network operator.

And there's the rub. No wonder telecom companies with big wireless exposure are seeing their stock prices fall. After paying some extraordinarily high prices for 3G licences, there is now a question mark over the size of their future revenue streams.

Bertelsmann e-commerce group CEO Andreas Schmidt calls it "the growing revenue gap". The good news for mobile service providers like Vodafone is that by 2003 85% of mobile phones will be internet-enabled. The bad news is that the high cost of the licences, added to the cost of building 3G networks, as well as putting in place the bridging GPRS technology, is putting big financial pressure on operators. These are costs that may not be matched by revenues. European telecom companies spent $95bn on 3G licences in 2000 and are expected to spend another $125bn over the next seven years to build base stations. 3G will not launch commercially until 2002. It will be at least five years before there is any return on investment.

"The landscape has changed," Schmidt told the Mobile Business Forum in Amsterdam last month. "We believed at the peak of the market that the service providers would dictate how this market would be constructed. But now they are having to partner with us."

So how did the operators get the numbers so wrong? The way that new mobile spectrum capacity was sold off certainly contributed to the current state of affairs. Getting its 3G auction out first made the UK Treasury a lot of money (some £22.5bn) but it also taught the bidders a lesson. Notice that the number of bidders fell in the ensuing auctions and the prices paid - after another hefty win for the German Treasury - came down rapidly.

Ironically, the UK government became a victim of its own spectrum sale success. When it came to selling off the fixed broadband licences for 28 GHz only a paltry £38.2m was raised compared with predictions of £1bn and 26 of the 42 licences went unsold. Telecom companies strapped for cash and unable to tap the jittery equity markets, or raise cheap debt on the back of concerns over mobile business models, held back.

Now that DoCoMo has signed equity deals in Europe with KPN and in the US with AT&T Wireless, the i-mode approach is on its way on both sides of the Atlantic. Just how it is rolled out remains to be seen, but it will most probably be as a hybrid technology with GPRS and eventually 3G. Meanwhile, Japan will roll out the world's first 3G services in May 2001. So far a lot of i-mode's success has been in offering subscribers the ability to download polyphonic ring tones and simple animations. The new 3G devices will also have games that commuting Japanese can play on their own, or with others.

Will this kind of service work in Europe and the US? This may be the multi-billion-dollar question. One thing is sure - the market is growing fast. Nokia estimates that there will be 200m internet-enabled handsets by the end of 2001. Perhaps the really tough question is: how many operators will be offering the services?