I was recently in LA, at the Barclays Tech conference in SF, and visiting portfolio companies in Silicon Valley. In LA, I visited a private company, Facecake, that could be on the verge of something big. Facecake's software allows users to try on various clothes and accessories online using their own image. Management claims to have an extensive user database with an average of 550 datapoints per user (this compares to Facebook' s Like feature with 50 to 75 datapoints). Its new Swivel 3D dressing room product and try-on banners are now catching the attention of some major companies according to management. If you are going to CES in Las Vegas, you may see some interesting demos from Facecake and its OEM sponsors.
The Barclays Tech Conference was a quality affair, with hundreds of top-line tech companies ranging from Microsoft (MSFT) to leading-edge big-data companies like VMware (VMW) and Fusion-io (FIO) to e-commerce investments like Ariba (ARBA) to mobile leaders like (QCOM) and a host of other semiconductor companies.
Cloud computing has been a theme that has struck a chord with many investors over the past few years. Cloud computing can be roughly broken down into two main categories: Big Data and Software as a Service (SaaS). Big Data replaces more expensive/less flexible in-house servers and storage with public and private out-sourced storage facilities (a.k.a. "clouds"). SaaS replaces more expensive/less flexible enterprise software with software subscription platforms that are hosted in the cloud.
Synonymous with Big Data is VMware. VMware's core business is server virtualization. Its value proposition has evolved from cost savings (by cramming more memory into its servers) a few years ago, to greater flexibility and productivity today. Its Vsphere operating system and desktop virtualization enables end users to utilize any device such as an Apple (AAPL) iPhone or iPad on a corporate network. The company's recent VRAM pricing model has encountered customer resistance by putting memory size limits for a certain price. Management has put-off the problem for now by raising the memory cap.
VMware recent move into desktop virtualization initially targets verticals like health care, federal government and finance companies. The key competitor in this space is Citrix (CTXS), whom we recently met in Geneva. Citrix XenApp and Xen Desktop have a more open architecture via paravirtualization than Vsphere, and can more easily interoperate with public cloud offerings like Amazon (AMZN) Web Services and Linux-based offerings from a host of cloud server providers. While VMware subsequently offered paravirtualization on its base operating system (a.k.a. hypervisor), Citrix's hypervisor is available on more platforms today than that of VMware.
Fusion-io's presentation was squatting-room-only, as there was no place left to stand. Fusion-io's success can be derived from Amdahl's Law, named after computer architect, Gene Amdahl. Amdahl's argument states that the speedup of a program using multiple processors in parallel computing is limited by the sequential fraction of the program. For example, if 95% of the program can be parallelized instead of serialized, the theoretical maximum speedup using parallel computing would be 20x. In algebraic form, the theorem is 1 / (1 - P). Thus, a rising portion of parallel processing (as opposed to serial) creates an accelerated rise in overall processing speed and efficiency.
Fusion-io has taken the flash memory trend, and maximized its efficiency by putting its non-volatile memory sub-system right below the application. This is opposed to competitors like EMC (EMC), who use a tiered approach with hundreds of instruction sets / emulations between physical memory and the native application. Thus, Fusion-io enables more parallel computing, and faster processing with less resources. Management claims density improvements of 3x to 6x today, which translate into lower total cost of ownership benefits. Revenues have skyrocketed since the June 2011 IPO, aided by its Web Scaling division where data center clients like Facebook make bulk purchases of capacity, along with strong growth in its core enterprise business. Management projects revenues to grow 55% in 2012, and are modeling long-term operating margins of 20% to 22% versus 8% last quarter.
Fusion-io stock price has fallen sharply since mid-November despite the strong growth outlook. Possible reasons include expiration of the IPO lock-up in December, lumpiness of its Web Scaling business, and a very rich valuation.
On the SaaS side, the conference was abuzz with the $3.4B acquisition of Success Factors (SFSF) by SAP (SAP) at $40 per share. Success Factors and peers such as Taleo (TLEO) offer several subscription suites of human resource and talent management to large enterprises at prices as low as $50 per seat. This modular approach saves corporations tons of money, as they only used what they needed. Moreover, productivity improved due to ease of use across the entire enterprise, with a much quicker implementation time than traditional enterprise software programs. Success Factors successfully gambled on top-line growth (+77% Q3 y-o-y sales rise) instead of profitability, to maximize its strategic value. Taleo, on the other hand, has managed to earn a small recurring profit, while still boosting the top-line 41% last quarter. Taleo's CFO stated that its outsourced talent management product suite (recruiting, performance management, learning, and compensation) still has low market penetration with a lot of room to grow.
E-Commerce can be broken down into two general categories: B2C and B2B. The B2C leaders are eBay (EBAY) and Amazon. Specific B2C category leaders include Groupon (GRPN) - online couponing, Netflix (NFLX) - online streaming of movies and TV shows, OpenTable (OPEN) - online restaurant reservations, Ancestry.com (ACOM) - online family history, Shutterfly (SFLY) - online photo sessions, Zynga (ZNGA) - online interactive games, among several other internet B2C categories.
The lesser-known area is the B2B online market, which has interesting potential. Ariba has emerged as the leader with its online business buyer/seller exchange network. Ariba's B2B network (a.k.a. Discovery and Quadrem) achieved commercial volume of $202B in the past year, 3x larger than its closest competitor Elemica (Chemicals vertical) with $60B, and doubling the $94B of transaction volume of B2C players Ebay and Amazon combined. Management feels its addressable commercial market size is at least $1 trillion (the entire commercial market is $12 trillion), so it is still early days.
Ariba's commercial exchange benefits from Metcalfe's Law or the network effect of each additional participant adding exponentially more connections and value. The number of registered members has increased from 289k in 2008 to 730k in 2011 due mostly to organic growth. Ariba facilitates this growth by being ERP-agnostic, so it can interoperate with Oracle (ORCL) and SAP, who also compete with Ariba. Ariba also continually adds functionality and more automation to make its network more robust. As Ariba's commercial network grows, so does its competitive advantage and strategic value to would-be acquirers like IBM (IBM), Oracle, and SAP.
Setting up supply chain markets takes work. While sellers are not too difficult to add, corporate buyers need to be educated with system tools. This "change management" process from paper to automated online purchasing precludes Ariba from the hyper-growth experienced by certain B2C peers like Groupon. On the other hand, Ariba has created consistent annual organic unit growth of about 25% in business network subscription revenues (mainly seller transaction charges) and near 20% in network applications (mainly buyer network modules). FY 2011 and FY 2012 network growth was higher due to acquisitions and a one-time price hike in FY 2011. These combined network Subscription Software (SaaS) revenues constitute 69% of total FY 2012 estimated sales of $525M, and 85% of the total client base. Fast-growing SaaS revenues have above-average gross margins of over 80%. Operating margins are expected to reach 25% in three years, versus 18% currently, with 15%-20% annual total revenue growth.
|Ariba, $,M, FY: Sep||FY11||FY12e||b-Process||FY12e total||total growth||prod mix|
|Total Subscription SW||276 |
|Consult & Training||71||83||4||87||+23%||17%|
|Legacy Client Svcs||38||28||28||-26%||5%|
|Total Services||109||111||4 |
|Optg Inc Non-GAAP||80||105||+31%|
|Optg Mgn Non-GAAP||18%||20% |
* organic growth +24% network, +19% total subs SW
FY11 GAAP reconciliation: stk base comp $58M, amort intang $13M, restruct $9M
FY12e GAAP reconciliation: stk based comp $70M, amort intang $20M
Various semiconductor companies echoed the same messages:
- Inventories are at all-time lows, but we still are not seeing a pick-up in demand.
- We are not quite at the bottom yet. We may still feel the effects of Thailand flood-related component shortages in early 2012.
- The euro uncertainty has delayed some end-customers' buying decisions.
- This is not as bad as 2008.
One sell-side analyst today put a buy on the semi equipment sector due to some rush orders from Samsung (OTC:SSNLF), while in the same tech sector overview, presented across-the-board Q4 earnings cuts from most semiconductor companies (the semi equipment customers ex-Samsung!), including one of the last holdouts - Intel (INTC). Sandisk (SNDK) may be the sole semiconductor stock remaining without a negative Q4 earnings pre-announcement. Sandisk is a beneficiary of strong NAND flash memory growth, especially from solid-state drives.
Cyprus Semiconductor (CY) said that business is soft overall, with its touch screen controllers performing relatively better. Cyprus' Gen 4 is the first touchable screen controller with a 32-bit ARM cortex processor. Management stated that Cyprus and Atmel (ATML) continue to be the touch market leaders, and are not losing much market share to new low-end competitors.
Teradyne (TER) stated that Q4 should be the bottom based on historical peak-to-trough declines. An experienced fund manager from Michigan then asked, what if demand just stays flat and does not recover in 2012. While semis are usually cyclical, below-average inventory levels could lead to this unprecedented flat trend. The bottom-line is that there is very little visibility today for most semiconductor companies.
That said, several semiconductor companies are running lean operations, leading to less losses or even profits at the bottom of the cycle. Teradyne, a historical money-loser in down-cycles, expects to report a 9% operating margin in a near-trough Q4 2011 quarter, and has averaged a 24% operating margin during the upturn of the last eight quarters.
Improved cost efficiencies are not exclusive to the semiconductor industry. Corporate America on average had record-high operating margins in Q3 2011. Most cost cutting initiatives have already been taken, however, making it difficult to sustain these margins without revenue growth. In fact, revenues should decline in the coming quarters given tepid global demand. The result should be continued downward earnings pressure for at least the first-half of 2012.