Originally posted by: Accord99
The article is one Itanium-unfriendly spin on the sales figure. A Itanium friendly one could go like this:
1) ~110,000 I2 processors were sold in 2003, a significant increase over the year before and many sales occurred after the introduction of Madison at the end of June.
2) I2 system sales have been increasing at a rapid rate quarter over quarter; IDC estimates are 2003 Q1: 1900, Q2: 3250, Q3: 4957, Q4: ~8900 and Intel expects to double I2 processor sales in 2004.
3) For all the talk of Opteron's momentum, the total number of Itanium 2 processors sold in 2003 were likely to be higher than the number of Opteron processors sold, generating much higher revenues and used in systems of much greater value. Indeed, AMD management were tight-lipped about comparisons of relative sales between Opteron and Itanium in the second half of 2003.
4) HP's revenues from its I2 servers increased by 60% in its most recent quarter compared to the previous year and helped revenues from servers exceeding $250K in cost increase by 9%, despite declining sales of Alpha and PA-RISC based systems
5) If Intel can sell 200,000 I2 processors, that should generate several hundred million dollars of revenue which should be enough to completely fund all research and development of IPF processors, chipsets and compilers.
A few other things to consider:
- Dean Takahashi, one of the authors of the article, is a well-known Sun cheerleader. Casting Itanium as a technology that customers don't want or need in the first paragraph is hardly a good start for a fair article. For crying out load, he quotes Sun's CEO Scott McNealy. He seems to be eager to get snappy one-liners from Itanium critics, but ignores other analysts and its current growth momentum. IDC predicts Itanium servers will bring in $7.5 billion by 2007 (about 20% of the market, and IMO a conservative estimate) and Insight64 believes Itanium will reach 2.5 million processor units/year in the near future (this predication came after the 64-bit Xeon announcement). Even Peter Glaskowsky of MDR, often an Itanium critic, had this to say after the 64-bit Xeon announcement:
"Itanium is the fastest, most scalable server-processor family now on the market, the result of billions of development dollars. Xeon can?t possibly be ready to take over Itanium?s role in the server market before 2007. By that time, Intel will probably have sold over a million Itanium processors, making Itanium a solid commercial success in the market for high-end servers."
- The press, being fond of a "David vs. Goliath" story, seems obsessed with Itanium vs. Opteron server unit numbers. While Opteron shipped in more servers in 2003, Itanium's processor unit shipments and server revenue were higher. IDC reported in Q3 that Itanium revenue was $123 million vs. $61 million for Opteron. If we extrapolate these numbers for 2003, Itanium server revenue was $472 million vs. $200 million for Opteron.
- Sun SPARC sells in 75,000 systems per quarter, IBM POWER in 25,000 systems per quarter, and HP PA-RISC in about 20,000 per quarter. Even if Itanium's current growth slows considerably, it will mostly likely be outselling PA-RISC and POWER by the end of the year, and SPARC in 2005 or 2006.
- Itanium is currently growing the most in the 4- and 8-way and up market, which commands about 10% of the server unit volume but over half of the server revenue. This is a space in which x86 has very little penetration...with Itanium replacing PA-RISC and SPARC shrinking by 15% per year, there is a lot of opportunity here.
- The article insists that Itanium is only doing well in scientific computing, but last fall Intel disclosed that Itanium is now seeing more sales to enterprise customers than scientific clusters. If you look at the enterprise benchmarks (TPC-C, SPECjbb, SAP SD, Oracle), Itanium currently has a 30-50% (or more) performance lead over Xeon and Opteron in enterprise applications, such as large databases, enterprise resource planning, supply-chain management, and business intelligence.
- At IDF, Intel disclosed that Itanium and Xeon will share the same platform in the future. The intention is that Itanium will have cost-parity with Xeon platforms while maintaining twice the performance for enterprise applications. This will go a LONG way to helping Itanium gain success in the lower-end 2-way market.
- Itanium development is not slowing down, and the upcoming products (I've worked on a future Itanium processor) are going to blow some socks off.
I can't claim to see into the future, but the game is by no means over (as the critics like to paint it), it is just starting.