This wasn’t a remarkably good earnings report – kind of bad, actually – and guidance for 1Q17 wasn’t exactly inspiring. Yet market reaction was unmistakably favorable, and that’s an understatement! What just happened here? My analysis will be long, but there’s a lot of important ground to cover. I eagerly await an Infinera earnings report where I have less to say. I’m sure you do too!
Earnings Report Highlights
The earnings press release can be found here: http://investors.infinera.com/new-releases/press-release-det…. Seeking Alpha’s transcript of the conference call can be found here: http://seekingalpha.com/article/4044714-infinera-infn-q4-201…. (Thanks, Seeking Alpha.) Unless stated otherwise, all italicized quotations are from the Seeking Alpha transcript.
4Q16 Revenue: $181.0 million Revenues declined 30% year-over-year and 2% sequentially. Management had guided towards $165-185 million and Wall Street expected $175.4. So we have revenue on the high side of guidance and beating market expectations. That’s nice, I suppose, until you notice that 4Q16 results were also 3% below pre-acquisition 4Q14.
In millions (GAAP – Generally-Accepted Accounting Principles). 1Q 2Q 3Q 4Q Comments 2012 104.7 93.5 112.2 128.0 2013 124.6 138.4 142.0 139.1 2014 142.8 165.4 173.6 186.3 2015 186.9 207.3 232.5 260.0 3Q15 Transmode acq. 2016 244.8 258.8 185.5 181.0
FY16 Revenue: $870.1 million 2016 featured strong first and second quarters and weaker third and fourth quarters, while 2015 featured the reverse. Comparing the two is basically a wash – 2016 was down less than 2% from 2015. Presumably, though, 2016 revenue should have been higher just because of the Transmode acquisition.
4Q16 Gross Margin: 38.1% GAAP and 41.8% non-GAAP Management had guided for non-GAAP gross margins of 40-42%. While it’s good that results were near the high end of guidance, these gross margins are well below the closer-to-50% margins seen even a couple of quarters ago. I’ll discuss reduced gross margins in a separate section.
4Q16 Operating Margin: -25.3% GAAP and -3.0% non-GAAP Guidance had been for non-GAAP operating margins of -10%. It’s hard to put a good face on negative margins, but the actual result was much better than guidance.
4Q16 Net Income: GAAP $-36.3 million ($-0.25 per diluted share) and non-GAAP $17.0 million ($-0.12 per diluted share) Management’s guidance for non-GAAP earnings per share was a range from $-0.16 to $-0.12, so they came in at the high end of guidance. Wall Street expected $-0.13, so Infinera beat expectations.
Earnings per Share (GAAP) 1Q 2Q 3Q 4Q Comments 2012 -0.19 -0.27 -0.17 -0.14 2013 -0.13 -0.09 0.03 -0.08 2014 -0.04 0.04 0.04 0.06 2015 0.09 0.13 0.06 0.08 2016 0.08 0.08 -0.08 -0.25 Earnings per Share (non-GAAP) 1Q 2Q 3Q 4Q Comments 2012 -0.07 -0.05 2013 -0.06 -0.01 0.10 -0.00 2014 0.03 0.11 0.11 0.13 2015 0.16 0.18 0.22 0.21 2016 0.19 0.22 0.05 -0.12
FY16 Net Income: GAAP $-23.9 million ($-0.17 per diluted share) and non-GAAP $49.4 million ($0.34 per diluted share) I think management is happy to see 2016 in the rear-view mirror, especially the second half.
4Q16 Cash Flow From Operations (CFFO): $-5 million; CapEx: $10 million. Negative CFFO and even more negative free cash flow (FCF: CFFO minus CapEx) is disturbing, but Infinera has a lot of cash on the books (~$345 million, excluding “restricted cash”). Management doesn’t expect negative CFFO to persist for many more quarters.
Revenue: $167-177 million (vs. $244.8 million in 1Q16)
Non-GAAP Gross Margin: 39-41% (vs. 50.2% in 1Q16)
Non-GAAP Operating Expenses: $89-93 million (vs. $92.9 million in 1Q16)
Non-GAAP operating margin: -13% at the midpoint (vs. 12.3% in 1Q16)
Non-GAAP earnings per share: $(0.18)-(0.14) (vs. $0.20 in 1Q16)
GAAP earnings per share: $0.13 less than non-GAAP earnings per share
Per Yahoo! Finance, Wall Street had previously expected 1Q17 revenues of $170.7 million and $-0.11 in non-GAAP earnings. Wall Street seemed willing to look past the missed earnings guidance.
Guidance Beyond 1Q17
Chief Financial Officer (CFO) Brad Feller doesn’t usually give much guidance beyond one quarter, but he offered a fair amount this time. Perhaps he recognized that a bad earnings report and bad guidance needed to be offset with talk of a better future. CFO Feller suggested a “… return to sequential quarterly revenue growth over the course of 2017.” That’s certainly better than the alternative, but first quarter guidance suggests a greatly diminished base from which to see sequential improvement. He further suggested that gross margins might not improve “… for the next couple of quarters as we invest to bridge existing customers to our new products, begin to deploy footprint with our next-gen solutions, and absorb fixed costs across a smaller unit base. Subsequently, my expectation is as we gain traction in the marketplace with our new portfolio of products, margins will begin to steadily climb.” SG&A (selling, general & administrative) and R&D (research and development) costs should remain elevated through 2017. However, my favorite part of post-Q1 guidance came during the Q&A.
Analyst: “… as I’m looking at 2017, is it reasonable to think we can at least get back to quarterly profitability in the back half…?”
CFO Feller: “Yes.”
I’m sure that CFO Feller is talking about non-GAAP profitability, which leaves about thirteen cents of room for error towards GAAP profitability. But I like the trajectory that statement implies, and the conviction in his voice as he answered.
INFN earnings day share price: $11.96 +27.23% (vs. S&P 500 +0.36%) I personally don’t believe that these results warranted a 27% upside move, but I’m also quite willing to grant that INFN was very undervalued in the single-digits. It may still be undervalued, but certainly less so. INFN shareholders have had a very good year or three these past few months!
Chief Executive Officer (CEO) Tom Fallon disclosed that the Cloud Xpress (CX) customer count increased by six in 4Q16, to 33, after not increasing in 3Q16.
CFO Feller returned to categorizing the top five customers in the quarter, which included “… two wholesale and enterprise carriers, two Tier 1s and an ICP” (Internet Content Provider). There were two 10% customers for the quarter: a wholesale and enterprise carrier and an ICP.
More unusual was the discussion of consolidation among Infinera’s customers, which included some telling metrics. CEO Fallon disclosed, “… our customer base is being consolidated at a fairly [ferocious] rate… About 40% of our revenue is currently in the process of being merged or acquired. … that’s a substantive amount of our revenue.” That is a big-deal statistic, and it makes me think that – ICE4 delays notwithstanding – Infinera must have executed quite well in 2016. Looking at 2016 versus 2015, revenue was down only 2%. Of course, the Transmode acquisition added another roughly 16% to Infinera’s revenue heft. With that transaction occurring late in the third quarter of 2015, Infinera’s 2016 revenues should have increased by ~11-12% over 2015 revenues, if old-Infinera’s and old-Transmode’s revenues were flat. In the face of a major slowdown from 40% of Infinera’s revenue-weighted customer base, an acquisition-adjusted decline of roughly 14% isn’t that bad. Also, CEO Fallon quickly added after that disclosure, “I don’t believe in any situation, except maybe one, we end up in a bad spot.” The “except maybe one” seems to refer to Verizon buying XO.
On a MUCH more positive note, CEO Fallon shared this, “… in Q4 we received an initial metro order from a large North America cable operator. While the ramp for this customer should be gradual, over time I expect this to develop into a sizeable opportunity. This deal represents a major milestone in the evolution of our metro business…”. CEO Fallon added later during the Q&A that a second Metro production order was received from this customer during Q1, quipping that his sales folk believe that a customer is still a prospect until after the second production order is placed.
Long-haul (and Subsea)
CEO Fallon didn’t have a lot to say about the long-haul market. He acknowledged that it still accounts for the majority of Infinera’s revenues. He further quoted industry research that long-haul would likely grow at a low-single-digit rate in 2017. However, he does expect bandwidth demands to force the long-haul market to increase its growth in coming years.
In the subsea portion of the long-haul market, CEO Fallon admitted to lost opportunities due to the delay of ICE4 (as they’re now calling the imminent generation of transport hardware), but noted that, “… we have multiple ICE4 subsea field trials with major customers being scheduled for the first half of the year.”
ICE4-based long-haul products aren’t expected to officially ship until the second half of 2017.
Data Center Interconnect (DCI)
Infinera saw DCI “… revenue more than double year-over-year …” in 2016. The market is still expected to grow in 2017 in the high-double-digits, which is pretty amazing. At this point, Infinera still has 90% market share. That’s down from 94% last quarter, but still a very strong market share.
One analyst with sharp pencils (and I mean that in a good way) suggested that DCI must be making up about 10% of Infinera’s annual revenue, which CFO Feller described as “… roughly correct …”. That’s a good number to have a feel for. Even if Infinera can only coax 50% growth in their revenues from the DCI market over the next several years, it won’t take too long before it becomes a significant percentage of revenues.
CEO Fallon noted, “… in January, we shipped our first Cloud Xpress 2 units to major ICPs for lab testing.” The earliest revenues are expected in the second quarter.
The complexion of the DCI market seems to be changing a bit, with the emergence of a “… rapidly-growing next tier of ICPs that are building their own DCI infrastructures.” CEO Fallon talked about several wins with this new tier of customer.
But CEO Fallon also hinted at the possibility of gaining another important ICP customer, although no names were mentioned: “… I firmly believe that we’ll be able to add a significant new ICP this year based on the CX2. We have to still earn it, but we’ve been given the nod for they will do the testing. … the hardest milestone for us historically is getting them to agree that they will do the testing, because they’ll only do the testing and sign a contract if they have a business driver and they believe it’s a viable solution for them. We have achieved that.” Alexa, which major ICP is still not an Infinera customer?
Metro Core, Aggregation, and Access, and Mobile Front- and Back-haul
CEO Fallon admitted that his timeline for Metro penetration was too optimistic. I don’t think that is a revelation for anyone who has stayed up-to-date with this company.
However, Infinera is starting to see some wins, and that’s very encouraging. The large NA cable operator mentioned earlier is a great example, but 2016 saw Infinera winning “… a combined 44 new XTM and XTC metro customers, ranging from smaller deals sold through our expanding global channel program to larger direct sales.” My guess is that many of these were indeed small, and occurred in the latter half of the year, otherwise Infinera would have had more to say about them earlier.
The Metro market is estimated to grow in 2017 at a high-single-digit rate, and hopefully Infinera will be able to further enhance that by grabbing market share.
An upgrade to the TM series products is expected this summer. CEO Fallon did not specifically say whether the upgrades were based on ICE4 or not. He did mention, though, that some customers want support for more modern modulation schemes, specifically mentioning 16QAM.
During the Q&A, CEO Fallon noted that Infinera has won several front-haul opportunities, mostly in Asia. That’s understandable because this architectural innovation was invented in China. Infinera has some good white papers about front-haul and back-haul on their website. Maybe I’ll try to digest them on this board someday, but this post is crazy long already! CEO Fallon sees mobile front- and back-haul as a “… substantive potential opportunity… ”, but it’s also likely to ramp up very slowly. In the U.S., it might get its best traction as part of a 5G roll-out, so we’re probably talking “next decade”.
Management offered virtually no commentary on software sales this quarter.
That said, management described three “paths” that Infinera will take in moving their technology forward. Although the path I’m about to discuss partially relates to hardware, my discussion of this path seems to fit most cleanly under a software heading. CEO Fallon talked about “… our commitment to lead in the open network philosophy that is becoming instrumental in the Layer T and C architecture”, and described this as having two main components. The first component is interoperability across different hardware vendors. That is, making sure that customers can use one vendor’s transponders and another vendor’s line system and have the combination work flawlessly and seamlessly. The second component is software-defined networking using APIs (application programming interfaces) that are open (i.e., non-proprietary; not specific to Infinera). Infinera believes that they offer exactly this with their new Xceed Software Suite. I’ll describe the other two technology “paths” in later sections of this report.
Transmode Follow-up (Probably the Last One)
Acquisitions, under Swedish law, work a little differently than in the U.S. Although Infinera’s tender offer for Transmode was successful, minority shareholders who didn’t tender their shares can remain Transmode shareholders for a certain period of time, until a “cramdown” process forces them out of their position. The cramdown occurred during Q4. The good news for Infinera is that it no longer needs to keep a separate set of Transmode books to preserve the rights of Transmode shareholders. The bad news for me is that I’ll no longer have a window into Transmode’s profitability as a separate entity. What I can say is that the 2016 loss attributable to Transmode shareholders was $0.5 million, or roughly 2% of Infinera’s total loss for the year. While that’s not encouraging, fourth quarter results may have shown an improvement, as CFO Feller described “… solid results from our traditional metro customers.” Since Transmode had been profitable before the acquisition, I wonder whether it was the addition of Infinera’s fledgling Metro products that hurt profitability.
Why Are Gross Margins so Weak?
In my opinion, there are two major factors, and two lesser factors, that are currently depressing gross margins. I would estimate that the impact of the major factors is roughly equal. The first major factor is Infinera-specific and the second is more industry-wide.
Infinera must allocate its largely-fixed manufacturing expenses – due to its vertical integration – over fewer sales when revenues are depressed. That this drives down gross margin shouldn’t be surprising.
Second, and more surprising, is this observation from industry analyst Dell’Oro: “… 100-gig ASPs [average selling prices] declined 29% in 2016.” When the unit price on a product declines 29%, simple math tells us it takes 41% more unit sales just to maintain flat revenues. What incredible headwinds Infinera was facing in 2016, when you combine this fact with the industry consolidation discussed earlier!
Year-over-year, the cost to produce a unit in Infinera’s foundry probably doesn’t change much. Therefore, a 29% decline in selling price will affect gross margins dramatically. Now, Infinera didn’t say that its ASPs declined by 29%, only that an industry analyst believes that prices declined by that much in the area of the market where Infinera focuses. Surely there was some impact to Infinera.
The first (presumably) lesser factor is assistance that Infinera is giving their customers. This was mentioned several times during the conference call, always couched in the term “investment”. Discussing lower margins, CFO Feller laid some blame on “several investments … to preserve existing business in anticipation of our next generation products …”, later adding “… we invest to bridge existing customers to our new products …”. There was no explanation forthcoming of exactly what this means or quantification of the cost, but it sounds to me as if Infinera is cutting deals to keep certain customers who are annoyed with the ICE4 delay.
The other (presumably) lesser factor CFO Feller mentioned is “product mix”. Historically, when Infinera was solely a long-haul business, this was a phrase that implied the differences between lower-margin sales that increased footprint (i.e., gaining new customers or increasing the installed equipment base at existing customers), and higher-margin sales that added bandwidth to already-installed equipment (i.e., either physically adding new line cards or electronically enabling Instant Bandwidth). Now that Infinera sells into multiple markets, I’m less certain that meaning is intact.
Perhaps disruptive to how I’ve ranked the impact of these various factors, CFO Feller indicated that the combined effect of my two “lesser” factors probably had a bigger impact than the loading of the foundry (i.e., my first factor).
Finally, giving us hope for better gross margins in the future, CFO Feller commented, “… obviously, the cost structure of the ICE4 products is much better than the current generation.” While I’m happy to hear this, there was no explanation of why that should be obvious, or any quantification of the improvement.
Infinera Needs to Speed Up its Next Generation Cycle
Back in early 2015, FoolishErik (now, CMF_FoolishErik) posted a series of ten “Chapters” over the course of almost six months, which combined to produce an amazing deep dive into Infinera’s business. If you subscribe to a newsletter or portfolio service that recommended Infinera, you have access to these posts (sorry, they’re not on the free site). I cannot recommend the series highly enough. It is a lot of work to get through, but you owe it to Erik to make the effort. After all, it was even more work for him to produce it! Thanks, Erik, that was awesome! As part of my prep work for this earnings season, I reread the series and one thing jumped out at me.
In Chapter 4: Early Products, he wrote, “Infinera’s first product was the DTN, and it was a smashing commercial success. Launched in 2004, it has been their flagship product for the past decade. Though sales peaked in 2008, they have been rewarded since then with a with a long, slowly declining tail of high-margin sales.”
This tells me two interesting things. First, the razor and blade model was certainly still working for Infinera in the prior decade. The “slowly declining tail of high-margin sales” is bandwidth expansion into existing chassis, in large part. Hopefully it is still at work. It seems to be – gross margins have generally been good until fairly recently. Second, that four-year product maturation cycle might be important.
In Chapter 5: Modern Products, FoolishErik noted that DTN-X was launched in September 2011.
It is difficult to know exactly when DTN-X sales peaked, but I suspect they have. It used to be easy to estimate DTN-X revenues when Infinera had fewer products. Then, Cloud Xpress, with its three different interfaces, became available in late 2014 through early 2015. Later, in April 2015, Infinera bought Transmode. Even if we can’t isolate DTN-X sales, we can say with certainty that Infinera’s total revenues peaked in 4Q15.
What can also be said with certainty is that the INFN share price peaked on August 17, 2015 at $24.72 (closing price) before declining to a recent low of $7.32 (again, closing price) on November 4, 2016.
If Cloud Xpress follows the same four-year product maturation cycle, sales would peak in 2018, maybe 2019. However, I feel as if Cloud Xpress will be different (i.e., a shorter product maturation cycle), for three reasons. First, the DCI market as a whole is growing much faster than the long-haul market was. Second, when Cloud Xpress 2 – based on ICE4 – launches in 2017, that probably resets the clock. Third, since it is a very different set of customers, I think a different, compressed product maturation cycle is probable.
In sum, it wouldn’t surprise me if Infinera realized somewhat early that the ICE4-based generation of hardware wouldn’t be available until later than they had hoped. This may have influenced the timing of their push into the Metro space, including the purchase of Transmode, as a way of juicing revenues that were likely to flag. So far, Metro has been a tougher nut to crack than management originally expected. While DCI is – to my mind – an unmitigated success story for Infinera, the market is still currently too small to offset the decline in long-haul. Add to that the fact that competitors have released hardware with greater capacity than Infinera’s (at least until ICE4-based hardware is available), and you can see how we got into this mess. In my view, though, Infinera has committed to exactly the right approaches to solving this problem: (1) reducing the time between one generation of PIC (photonic integrated circuit) and the next; and (2) diversifying into adjacent markets where the product cycles are likely on a different timeline and cadence. We need to hold Infinera accountable for achieving these goals.
To underscore the importance of this observation, management has disclosed that they are committing to a two-year product upgrade cycle. They also disclosed that they have 600 Gb per wave PIC technology working (and being demonstrated) in their labs today. This improved product upgrade cadence should keep Infinera well ahead of the kind of product maturation cycle I described, and should help them stay ahead of competitors as well (although those competitors are now “on notice”). In fact, “increased product upgrade cadence” was the first technology “path” that CEO Fallon described, although I chose to mention it second.
Other Random Musings
Humorous Moment During the Earnings Conference Call
OK… You’ve made it this far through a pretty dull post. Time for a moment of levity…
Analyst: “… I just felt like maybe you’re a little bit ahead of what you thought you’d do in the Analyst Day.”
CEO Fallon: “I never want my engineers to hear that, Rod. I consider us distinctly behind where I want us to be.”
Multiple Voices: Laughter.
New Breakeven Run Rate
Before the Transmode acquisition, I had described Infinera’s breakeven point as $150 million in quarterly revenue. That figure is as high as it is because Infinera is vertically integrated – they own much of their own manufacturing capacity, which creates large fixed costs regardless of how many units are sold (please note: I am calculating breakeven based on GAAP expenses). Based on recent quarters’ financial statements, it appears that – post-acquisition – breakeven has increased to roughly $225 million in quarterly revenue. While it is perfectly understandable that the acquisition would raise the breakeven point, I am confused by the magnitude of the increase. Based on financial statements of both companies prior to the acquisition, it appears that Transmode’s revenues were roughly one-sixth the size of Infinera’s, and both companies were similarly profitable. Yet breakeven appears to have risen by roughly 50%. What is happening? I haven’t quite figured it out yet, but I’m suspicious that the impact of reduced ASPs on gross margins is no small part of the equation. If that’s the case, then all may not be lost! Infinera should be able to charge premium prices for ICE4 products, boosting gross margins. It could also be that Transmode’s recent lack of profitability (as described earlier) is hurting the breakeven point. Although we’ll no longer see Transmode as a stand-alone entity, improvement in Infinera’s Metro business could help matters. $225 million strikes me as high, but it seems to be accurate, whatever the cause.
Q1 is typically Infinera’s weakest quarter, seasonally. It’s the same for the entire industry. But for Infinera, it’s about to become weaker. A particular cable operator – one currently being merged or acquired – used to typically place its entire yearly order in Q1, with Infinera’s revenue recognition usually split between Q1 and Q2. The new combined entity has informed Infinera that it should expect orders going forward to be more evenly spread across the entire year. This is a little disappointing, as it was nice to have a customer offsetting seasonal weakness. But it is certainly not the worst possible post-acquisition outcome.
Research & Development as a Percentage of Revenues
CFO Feller indicated 2017 R&D expenses as a percentage of revenues would likely be in the high-20% range. This is well above long-time guidance of capping R&D spending at 20% of revenues, and also the mid- to high-20% limit described at Insight Infinera. CFO Feller did make two clarifications to this, though, during the Q&A. He commented that this really wasn’t a change in either R&D or revenue expectations, but instead a more accurate refinement of what expenses will be. He also reiterated that Infinera’s long-term goal is to bring R&D back down to 20% of revenues, but hopefully at higher revenue levels.
Build or Buy – Round 1
In answer to an analyst question (the same guy with the sharp pencils), Infinera co-founder and President Dave Welch disclosed that it buys some of its components externally and doesn’t manufacture them all in its foundry (and he mentioned that they don’t disclose who their partners are). I don’t find this disclosure surprising – some things are cheaper to buy than to figure out how to make. What I found surprising was that Mr. Welch made the disclosure at all in response to the question. The question, it seemed to me, was basically: “You’ve talked about your next generation PIC, but you haven’t talked about a next generation DSP (digital signal processor). When can we expect that?” However, the question did specifically mention Acacia, Ciena, and an Inphi acquisition (presumably ClariPhy Communications). Perhaps one of these names (ClariPhy?) touched a raw nerve.
By the way, the answer to the question is that ICE4 includes an upgraded DSP.
Build or Buy – Round 2
In recent months, we’ve seen reports that some of the major ICPs are seriously questioning their hardware purchases, with an eye towards potentially building their own solutions in house. So far, these have been specific to switches and routers; Infinera’s stronghold in the transport layer has been spared to date. The fact that Infinera is the only company that has managed large-scale PIC integration and production will likely delay customer encroachment for a good while longer. Here is one example: http://www.telecomasia.net/content/equinix-deploys-facebook-… (Thanks, SJ57.)
Integrated v. Disaggregated
I promised three “technology paths”, and this is the third that CEO Fallon highlighted. My takeaway from his comments was that service providers have historically required an integrated architecture where all components played well together, while ICPs preferred disaggregated solutions, where each piece of hardware was chosen to optimize a point-to-point solution. While the distinction may be blurring, CEO Fallon wanted to assure us that Infinera would provide great solutions for both architectures.
At several points during the call, demand for ICE4 products was described as “pent-up”. This is obviously encouraging, but most revenues are likely more than one quarter out.
I’m further encouraged that Infinera finally seems to be gaining traction in Metro, as evidenced both by many small wins, as well as that major NA cable operator. The potential for capturing the DCI business of another large ICP is similarly enticing.
As I stated earlier, I think the price rise that followed the earnings report was: (1) based on future prospects rather than near-term result expectation, which is unusual for Wall Street; and (2) at least partially due to bringing INFN up from severely undervalued levels to something closer to fair value.
I’ll end here because this is so long already. If you have any questions or comments about this post, please do ask/share them. If you found this post helpful, please recommend it, which helps others find it. If you value the work that I do putting these posts together, please click the smiley-face to the right of my screen name, making me a “Favorite Fool”. That helps me, and helps others find me and my work. Please do this for anyone whose work you value.
Thanks and best wishes,
TMFDatabaseBob (long: INFN)
See my holdings here: http://my.fool.com/profile/TMFDatabasebob/info.aspx
Peace on Earth