Peerstone Software Pundit

The IT Investment Tinderbox

A major piece on the front page of today’s Wall Street Journal (subscription required) argues that American companies are sitting on a huge cash hoard, but are extremely reluctant to spend it on new gear or software.

After declining in 2001 and 2002, investment spending finally turned up in the spring of 2003 and has been increasing since then. Shipments of U.S.-made machines, computers, furniture, airplanes and other capital goods in the first eight months of this year were 12.1% above year-ago levels…[snip]… But business isn't spending nearly as much as swollen profits and the pile of cash in corporate coffers suggest it should…

According to the Journal, US companies held $1.27 trillion of liquid assets at the end of Q2, up more than $300 billion since 2002. At nearly 11% of GDP, this is the highest rate of corporate cash holding since 1959.

Well then, it looks like someone has a little money to spend. The 64 terabyte question is: will they or won’t they? And if they do, when? The Journal story doesn’t really have an answer to these questions, but it deploys a few carefully chosen anecdotes to suggest that large US manufacturing companies are skittish about capital spending. Among the reasons cited:

• current capacity usage that is still not high enough to justify more than maintenance spending
  
• an economic recovery which is strong but not roaring
  
• more stringent financial reporting requirements
  
• excess capital spending in the last cycle
  
• new lean manufacturing technology that lets companies produce more with less equipment
  
• a preference for investing in overseas rather than US-based plants
  

This pot pourri of reasons not to spend is interesting. But it doesn’t add up to a causal mechanism that explains what we would characterize not as the softness, but rather the timidity in current enterprise IT spending. The Journal article goes on to argue that if consumer spending weakens in Q4, as more anecdotal evidence and a few non-committal economists suggest it might, then we’ll really be in trouble, because it doesn’t look to the story’s authors like business is ready to pick up the slack. Here we get to the real reason for this article’s placement on the front page, which is to hint that the Bush recovery might not be as solid as the Larry Kudlows of the world imagine (yes Virginia, the Journal’s news pages are decidedly more liberal in their slant than its editorial page).

But if we tune out the political subtext as irrelevant, it remains that the Journal has put its finger on a fact of extraordinary consequence: the current cash rich status of US companies constitutes a veritable tinderbox of potential IT spending that is awaiting the appropriate spark to set it off. Make no mistake: the tinderbox will not combust spontaneously. All other things being equal, executives who have survived the ups and downs of the past five years will choose to sit on their cash rather than spend it (or else distribute it to shareholders through buy-backs and dividends, like Cisco and Microsoft). Not unless some compelling outside influence forces their hand. The current situation can be compared to a dry forest where a vast amount of underbrush has accumulated that could provide fuel for a raging forest fire – if the right stroke of lightening came along.

What might that stroke be? For the past two years our survey research has shown that many purchasers of big packaged apps such as ERP or CRM, while acknowledging that the software was necessary, are disappointed in the hard dollar ROI produced by it. The fact is we are very late in the current software technology cycle. Most software vendors and venture capitalists have convinced themselves that today’s enterprise IT customer is only interested in incremental improvements that show quick payback. This is of course a self-fulfilling prophecy: if your big bang products fail to deliver big bang benefits (due to rigid architectures that encourage out-of-control labor costs and unmanageable integration complexity), then naturally your customers will fall back on small bang products that are less likely to disappoint.

But there are some exceptions. Oracle, PeopleSoft and Siebel, for example, are all looking enviously at the license revenue growth spurred by SAP’s Netweaver and MySAP offerings. We can only hope that the executives of the first three companies (who have been a little distracted lately, it is true, by M&A and management issues) are pressing their product development teams to match SAP. To be sure, if we look beyond applications to infrastructure, we see that Oracle’s own 10g database is doing rather well too – surely matter for reflection right now at IBM, considering the flat to negative sales growth posted by DB2 in the most recent quarter.

We don’t know exactly how the next IT spending forest fire will start. But we can describe the chain of events that will lead to the conflagration. First, a handful of prominent user companies will suddenly find they are getting major traction with some of the newer software technologies. Perhaps it will be SOA (services oriented architecture), RFID, datacenter virtualization, or something else. Second, sales results for the companies who supply these products – be they established giants or start-ups – will pick up, drawing the attention of Wall Street and the media. Finally, the executive peer group of the early adopter companies will notice what is happening and begin to wonder if they are being left behind. This last stage is crucial, for we believe that formal or informal peer group benchmarking is an extraordinarily important factor in driving big ticket capital spending decisions. Peer pressure is the wind that can turn a small fire into a big fire, one that jumps fire breaks and sweeps all before it.

Bottom line: the Journal’s skepticism notwithstanding, we think the odds favor a major new flare-up of IT investment that could begin in earnest as early as 2005, following what we believe will be a surprisingly strong final quarter in 2004.

What Conway's firing really means

So Craig Conway is finally out. What a long strange trip it’s been since the day Conway announced the JD Edwards acquisition in June of last year. In some ways these fifteen months have been a time of suspended animation for PeopleSoft. While SAP was roaring ahead with its vastly ambitious Netweaver strategy and Oracle was pushing into a major refresh of its flagship database line with 10g – both of which have produced exciting growth in new software licenses – remarkably little of note came out of PeopleSoft, except for the incessant bulletins from the antitrust and poison pill wars.

And therein lies, in our view, the real reason for Conway’s abrupt dismissal by the man who hired him, PeopleSoft founder David Duffield (technically Conway was fired by the board’s outside directors, but they would never have made such a move without Duffield’s at least tacit consent and agreement to resume the CEO role).

PeopleSoft-IBM: waiting for the beef

We predicted the PeopleSoft-IBM middleware alliance back in July in one of our Research Notes published to the Wall Street crowd on First Call:

PeopleSoft could attempt to accelerate the modernization of its multiple lines of application middleware by engaging in a strategic platform alliance with IBM, possibly sealed by a minority equity investment from Big Blue. The core of such a deal would be a multi-year redeployment of PeopleSoft’s application portfolio to WebSphere J2EE infrastructure, similar to what Siebel is now undertaking for Siebel 8. It would also entail a push to standardize DB2 as the preferred database for PeopleSoft customers (virtually ousting Oracle from new deals, while leaving a secondary role for Microsoft SQL Server) and increased promotion for PeopleSoft’s existing but little noticed Linux versions. It could even offer customers the option of using the IBM-sponsored open source Eclipse development environment as a tool for customizing PeopleSoft apps or building new add-on modules.

But now that the announcement has been made, we have to ask: where’s the beef? The statements issued by PeopleSoft and IBM so far are extremely short on substance, which is a sure sign that the two sides have not settled on anything like a detailed road map for their alliance, but have agreed to hold hands in public for a while in order to buy time. IBM lost little time distancing itself from Craig Conway’s claim that the deal would involve a joint investment of $1 billion.

Careless observers might think the deal means all of PeopleSoft’s applications will be ported to IBM’s WebSphere. Although that’s certainly a long term possibility, nothing suggests it will happen any time soon. PeopleSoft’s flagship Enterprise portfolio is a byzantine patchwork of modules built over many years using the vendor’s own development environment and runtime middleware, PeopleCode and PeopleTools. It also incorporates assorted third party components such as BEA’s Tuxedo transaction monitor as well as a web page server, which can be BEA WebLogic or IBM WebSphere. Most of the installed base runs on Oracle, not IBM DB2. Add to this two other, completely different ERP product lines inherited from JD Edwards, one built in the proprietary IBM OS/400 environment (World) and the other built on a traditional Unix-Windows client-server architecture (EnterpriseOne). Porting all this to WebSphere and J2EE would be a gigantic undertaking, far too risky and expensive to contemplate in PeopleSoft’s current embattled circumstances.

What the announcement really boils down to is something far more limited, but nonetheless interesting. As we read it, PeopleSoft’s true aim is to deploy WebSphere in parallel to its own older middleware to respond to the roaring success of SAP’s Netweaver integration stack. Our survey work over the past few months shows that the SAP installed base is much more active at present than the PeopleSoft installed base. And the reason is Netweaver. SAP customers are not only migrating slowly but surely from R/3 to MySAP, which requires the Netweaver platform. They are also using it to tie their sprawling and chaotic non-SAP app portfolios into the ERP backbone or into the ERP-based data warehouse. PeopleSoft’s IBM alliance is pitched at large accounts that may be running both SAP and PeopleSoft apps, and that – absent a credible middleware strategy from PeopleSoft – are increasingly tempted by SAP’s siren call of “come integrate with me”. PeopleSoft’s argument seems to be: if you’re going to put all your integration eggs in one middleware basket, better it be IBM’s than SAP’s.

Bottom line: PeopleSoft’s IBM alliance is an expedient fix to the app vendor’s fear of being pushed out of complex large account deals by the superior breadth of SAP’s offering. We don’t think customers will see a whole lot of new software coming out of the deal any time soon. But it could be a compelling talking point for PeopleSoft sales reps to take into big-ticket RFP negotiations.

Meanwhile, despite the recurrent but silly “IBM as white knight” rumor which we believe is utterly unfounded (IBM is not going to wreck its far larger relationship with SAP by bidding for PeopleSoft against Oracle), the WebSphere alliance does nothing to solve Craig Conway’s most pressing problem, which is called Larry Ellison. We think there are one or two major surprises that Conway may have up his sleeve. More on that in a later post.

PS: Another obvious but little remarked upon consequence of the deal is to put both BEA and Microsoft squarely in their place as also-rans in the enterprise application middleware sweepstakes. There are now only two contenders who count: IBM and SAP.

Craig to Larry: it's not personal

Is Craig Conway softening? The SF Chronicle drew that conclusion in a weekend profile of the embattled PeopleSoft CEO, based on the following quote:

"My view is almost exclusively business and zero percent personal... [snip]... There isn't really the personal battle between Craig Conway and Larry Ellison that the media or even the common person would assume there is."

Craig Conway is no common person, but he is certainly a rational player. If he ultimately concludes that selling to Oracle is the best move he can make, he will make it. But we don't believe for a minute that he is ready to throw in the towel just yet, even if the rumored likely decision of the EC not to object to the deal is another body blow after the loss of the DoJ case (see our long piece on the extraordinary weakness of the DoJ case, which earned us a lot of yelling and screaming from PeopleSoft execs when we published it last June).

All the public statements at last week's big PeopleSoft customer shindig seem to telegraph continued resistance. A leopard doesn't change its spots so quickly. Conway’s kind remarks about Larry merely show his board and his shareholders that he hasn't made the effort into a personal crusade.

No, the game isn’t over. The only real question is whether the next move will be a battle about price, or whether Conway still has a game-changing trick up his sleeve. More on this subject later in the week.

How much does Microsoft discount?

Finding out what enterprise software really costs is usually a huge challenge for customers. Vendors go to extraordinary lengths to conceal the truth about their pricing. They don't disclose it on their web sites, in their press releases, or in their quarterly conference calls to Wall Street (though sometimes these calls might include limited but useful info about average selling prices, or ASPs). The reason for this reticence is clear enough. The incremental cost to manufacture another software CD (or enable a download) is close to zero. Everything is value add. Which means gross margins are astronomical, typically above 90%. Which means the software vendor's room for negotiation is... dangerously large. The Department of Justice's disastrous case against Oracle at least had the merit of exposing some real world software pricing data to the cruel light of day. We'll take a look at this data in a later post, but suffice it to say that what it says about pricing exceeds your wildest imagination.

A few vendors actually publish their list prices. This is more common for consumer than for enterprise products, but there are a few notable exceptions, such as Microsoft's Windows Server price list and Oracle's ERP price list. What is really rare is for a vendor to post their actual discounts off list price on their web site. But it seems Microsoft has taken this unusual step, as part of its relentless battle against Linux. On its Get the Facts site Microsoft posts a study apparently commissioned from Bearing Point which arrives at the surprising conclusion that Windows Server is actually cheaper to buy than Red Hat Linux! No need to go into all the gory details here. But a key assumption in the Bearing Point piece is that you can routinely buy Windows Server 2003 Enterprise Edition at a substantial discount. In fact, one of the tables in the study shows that a company buying just three licenses for the high-end server OS could expect to get a 49% discount off Microsoft’s published price ($13,784 paid over 5 years as a subscription vs. list price of $29,993, excluding support costs and client licenses).

This is great news for Microsoft customers, though maybe not for Microsoft resellers. We wonder if they really meant to publish it. But even if it was a Freudian slip, such rare transparency is priceless in the exceedingly opaque domain of software pricing.

Palmisano: a closer, not a philosopher

Palmisano's sales-driven personality emerges clearly in this interesting profile. Though he majored in history at Johns Hopkins, he built his career at IBM as an affable closer, not an intellectual. It's not surprising that his big idea - the controversial claim that a company the size of IBM can grow faster than the industry as a whole - is a salesman's idea. Lou Gerstner had already identified the source of IBM's future growth in services and outsourcing - aka business process transformation services. What Palmisano has to do is sell that idea, not only to his customers (we know he's good at that), but also to his 300,000+ employees.