I'm surprised that such an interesting question has raised no responses. Let me fix this by adding my own two cents, which is about 15 fen in China & 80 paise in India. My comments relate to the better class of analog IP providers, & I know some seriously good ones.
I see three segments in such analog IP:
1. Low-end: Crystal-less clocks, generic ADC/DAC, temperature sensors...
2. Middleware: Bandgaps, freq synthesizers, PLL's, USB I/O...
3. High-end: High-perf ADC (pipelined, Σ-Δ), RF signal chains, Serial...
There are two flavors of target customers: (i) digital IC designers who need some minor analog functionality but don't want to invest in in-house design, and (ii) analog designers themselves, under schedule pressure.
The first type of customer represents low-hanging fruit for the analog IP providers. It's a good steady source of revenue. The second type invariably wants high-end & represents revenue upside, and is also the most difficult to satisfy. To quote an Indian proverb (without prejudice) "It takes one snake to know another". In 90 nm or 65 nm the most important issues for this high-end market would be:
1. Availability of schematics along with GDS-II. Analog IP without schematics is useless.
2. Post-silicon design validation in the target process, preferably full PVT.
3. DFT, yield models, proven ATE test plan & execution.
4. Legal indemnification for IP issues.
5. Escrow arrangement as required (may be an issue for big companies)
Items 2 & 3 are the most important for creating real value. A good portion of the IP provider's resources will eventually get sucked into this phase.
Item 4 may be viewed by some as a serious show-stopper, but remember this is a business ecosystem. For example, if the customer is large they may have their own extensive IP portfolio and may prefer to utilize the analog IP provider purely for services. It is well known in the analog IP business that confrontation with your large customers is a slippery slope at best.
A final observation: in India & China analog IP companies are started by people who have substantial *local* industry experience, typically working for a transnational company. They have extensive knowledge of the local employment scene and may well offer a more challenging & rewarding employment track for designers in the established companies. In the end, it comes down to how many customers buy the same design - this is the ultimate commoditization of engineering where everything becomes a piece of software.
Your next <musicPlayer> may well carry a notice that its IC's were not designed in sweatshops because the designers get great stock options and work in air-conditioned offices
M.G.Rajan