While Intel has its separate thread, here info and discussion of other companies producing in the 100G space
From User 47433978 (see here for original)
I do not have channel information you ask for about Intel, but the following competitor analysis I address to you in this SA post on August 25 is still fundamentally intact now:
"Allow me to add some granularity on your comprehensive competition analysis on 100G QSFP28 CWDM4, 2km :
1. 100G cwdm4 producer also include ColorChip, Source Photonics. ColorChip uses a DML laser, contract FN for manufacturing. Source Photonics uses EML laser. Visit websites FFI.
2. InnoLights buys DML laser outside. A well-organized Shanghai CM in China, back away from Wall Street IPO in 2015 after investment from Google.
3. LITE, FNSR, OCLR all use EML lasers, and generate good sales in 100G QSFP L4 (10km reach). But EML laser, requires cooling is costly for cwdm4(2km reach), so companies using EML laser for cwdm products are not competitive as AAOI uses DFB DML type non-cooled laser. Visit each company catalogue FFI.
4. Luxtera and INTC trying to market SiPH cwdm4 products. Luxtera now uses an external laser on SiPH PIC, same process as AAOI. INTC cwdm product needs to overcome insertion loss and temperature issues from what I read.
5. I find no information as how successful INTC on its Hybrid Silicon Technology in making the QSFP type CWDM products. My opinion is if SiPH technology is not developed fully to on-chip or on-package stage, Intel cannot compete in the QSFP and OBO type product format after additional packaging cost and third party CM cost add up. INTC to become a dark horse could be 10 years away, sooner if they can achieve the die level integration of switching IC and optics IC.ie do without port plugging.
6. So from the above information, I have confidence in AAOI’s self-claiming, the lowest cost producer for 100G QSFP28 CWDM4 transceivers at least for now."
I also addressed at this board on August 24 the following,
" My view is when they progress to the stage of putting optics on-package and on-processor, the products will be IC foundry manufactured and will be disruptive to current FO companies including AAOI. My view is that day won't come in 10 years. The current At-board edge product (QSFPx) including 100G,200G,400G, or even 800G can play up to 2023 for up to 25.6TB rack density. At higher than 25.6TB, the high count foot prints (I/O) of switching IC is the limit for QSFPx products and the on-board product (OBO) will overlap and thrive up to 2017. AAOI can play well in these two form factors, QSFPx, and OBO and they have shown products in their technology roadmap."
SiPH is nothing to be afraid of. AAOI already has SiPh component in existing CWDM4 products and SiPh based long reach products in 400G technology map. AAOI has now two locations, i.e. US and Taiwan with SiPH products, presumably designed in-house and made by local merchant IC foundries. The 6" or 8" wafer CMOS foundry is fully depreciated for Taiwanese merchant foundry and readily accessible for AAOI assembly site in Taiwan. Note that Taiwan is the CM capital of electronic assembly primarily for US giants like Apple, Tesla, Nvda etc. I believe AAOI has inherent edge over its US peers in managing the supply chain cost due to its long time working relationship with local industry in Taiwan.
So I believe AAOI is still the lowest cost producer for 100G QSFP cwdm4 as of now. If AWS buying elsewhere in 3Q or after, AWS is paying premium or the seller is sacrificing GP to gain entrance, not a long term lasting strategy.
My model shows EPS2017 for AAOI is $4.99. I base nonAWS revenue will grow $19 million in Q4 to $83 million, AWS remains@$9 million.
Note that cwdm4 laser is still in short supply. CM like InnoLight gaining market share even with higher cost buying laser from OCLR tells why Non-AWS customers grab AAOI CWDM products with revenues increased notably in Q3.
PCL Technologies, a CM in Taiwan for Cisco, AVGO trying 100G QSFP cwdm on market by Q4 for customers in US. The Laser supply: VCSEL from Broadcom, 25G DFB from Melco, EML from Sumitomo. Like InnoLight depending EML laser from OCLR, PCL get lasers from outside sources. More cost to build transceivers. The Taiwan stock of PCL enjoys PE at 22. While AAOI's PE only < 10, the US market now is depressing the AAOI valuation down to unreasonably low point. Who is smarter investor? Time will tell!
User 47433978