Description

This is my favorite Short Sale idea, and apologies in advance, a longwinded write-up.

a) Brief Description and History: Cray Inc., formerly known as Tera Computer Company (“Tera”), is a manufacturer and servicer of vector-based supercomputers. Tera purchased Cray from Silicon Graphics in early 2000 for $50 million and renamed itself Cray. Tera was founded by Jim Rottsolk and Burton Smith to design a new supercomputer architecture called MTA. Cray stock is trading at approximately $7.20 / share (with at least 78.0 million fully diluted shares outstanding) and has a market capitalization of approximately $560 million, valuing the company at 27x the Street's average eps estimate for '03 and over 50X last year's "reported EPS."

b) I believe Cray will badly miss projections over the next two years. Cray management has touted in its recent equity road show the enormous opportunity that it has at Oak Ridge National Laboratory to build the fastest supercomputer in the world to rival a Japanese computer called the Earth Simulator. The project, which is contingent on securing funding from the government, is a major driver of improved results in 2004 and the foundation of the Cray growth story. In late 2002, while giving guidance to research analysts involved in the equity offering, CEO Jim Rottsolk projected that Cray would sell $200 million of X1 systems in 2004, $100 million to “regular customers” and $100 million to Oak Ridge. Since those meetings, guidance has not been revised even though in January, the President released his proposed 2004 budget that did not include funding for such a project for Oak Ridge. In fact, Oak Ridge funding for supercomputing in FY 2004 was reduced by 5.4%, a second consecutive year-over-year funding decline ($10 million in total for supercomputing).

Based on guidance from Cray management, sell-side research analysts expect Cray to generate 2004 Revenues in a range of $300 – $350 million, an improvement of 33% - 55% over projected 2003 results. Cray’s guidance identified a potential target market for its computers in 2004 of $640 million (including a $42mm “no” margin project for the Sandia National Laboratory). Included in the $640 million target market is $230 million for Oak Ridge. Absent this $230 million, Cray would have to win 65-75% of the remaining target market from rivals including NEC, IBM, DELL and SGI to meet its revenue projects. According to a senior Cray manager no longer with the company, Cray has lost 8 out of the last 9 bids for international weather projects as the company has difficulty competing against rivals on cost and ease of use as the competition has a more robust stable of 3rd party software and applications. With its competitive disadvantage in the commercial and international markets, securing such a large percentage of their identified target market would appear to be very difficult.

Project Amount
Target Market $642.3
Less: Oak Ridge (230.0)
Remaining Target Market 412.3
Amount Under Contract (42.0)
Market for Competitive Bids 370.0
Required Wins to Meet Projections 238.0 - 278.0
Required % Success Rate 64% - 75%

The following chart outlines the historical and 2004 budget allocation for Oak Ridge compared to other research institutions:

Laboratory 2002 2003 2004 % Increase
Albuquerque $9.5 $8.9 $9.6 +7.9%
Chicago $34.9 $18.0 $39.4 +118.3%
Oakland $72.8 $54.4 $64.5 +18.5%
Oak Ridge $27.0 $10.6 $10.0 -5.4%
Richland $4.1 $1.0 $3.6 +259.0%
Washington $1.8 $73.6 $46.4 -37.0%
Total $150.2 $166.6 $173.5 +4.2%
Source: FY 2004 Congressional Budget – Science / Advanced Scientific Computer Research – Page 389

Of the six laboratories that received funding for supercomputers, only two, Oak Ridge and Washington received reduced budgets. The two with the largest budgetary increases, Oakland (Berkley) and Chicago (Argonne) have publicly displayed their support of IBM-based (rather than Cray) solutions for scientific supercomputing.

c) Cray's accounting techniques have made profitibility appear higher and smoother than I believe it really was. In 2002, Cray recorded $5.4 million in net income (or $0.09 per share). This net income was augmented by one-time sales of written-down inventory, reversal of warranty reserves, reversal of purchase commitment reserves and other one-time adjustments. Eliminating these one-time adjustments, net income would be negative $5.0 (or –$0.09 per share), a variation of $10.3 million from reported earnings. In 1999, Cray recorded a one-time reserve of $32.3 million for warranty costs associated with its T90 product. The use of this additional reserve boosted net income by $5.8 million (or $0.11 per share) in 2002. Absent the use of this reserve, Cray would have generated net income of negative $10.7 million in 2002 (-$0.20 per share), a variation of $16.1 million from released earnings (or -$0.29 per share).

Looking at the reserve activity it appears to me that 2002 reported earnings may have been engineered to meet Cray's stated goal of being profitable in each quarter of the year. If you think about their business, continuously generating smooth quarterly EPS numbers should be inherently difficult as revenues are driven by very large contracts and any slight delay in a contract would have a significantly negative effect on quarterly operating profits. To note: on the March 5, 2003 conference call Jim Rottsolk, discussing the positive performance of Cray in 2002, highlighted the fact that Cray had been “profitable in all four quarters of 2002”. Read the 10q's and judge for yourself.

d) Corporate governance has been poor, with a former board member becoming enriched at the expense of shareholders. Notice the history of Cray financings involving Terren Peizer, the Chairman of the Board from 1999 – 2000 and Cray director through February 2002. Peizer was never elected by shareholders and was appointed as Chairman of the Board following his involvement in a 1999 Tera financing (of $31 million at $4.50 per share). As remuneration for his role, he received $650K in cash and 10% of the company in cashless exercisable warrants (with dilution protection that ultimately resulted in 5.2 million warrants with an exercise price of $2.53). Based on the current stock price, these warrants are now worth approximately $25 million. This fee represents a significant percentage of the current enterprise value of Cray. Subsequent to the 1999 financing, Peizer received an additional $650K finders fee for a $9.3 million debt financing in 2001 (7.0% fee). In February 2002, Peizer received a $273K finders fee for a financing that raised $3.9 million (7.0%). Following the completion of this financing, Peizer resigned from the Board of Directors. For trivia buffs, Piezer sat adjacent to Michael Milken in Drexel’s Bevery Hills high yield desk and was a key turncoat in Milken’s prosecution.

e) There has been significant management turnover. In 2002, Cray replaced both its CEO (after five months) and CFO.

Catalyst

Missed estimates, greater scrutiny of investors into accounting choices and corporate governance policies.