I had the chance to talk last night with Marvell CEO Matt Murphy and optics leader Inphi CEO Dr. Ford Tamer on its deal valued at $10B, which considerably expands Marvell’s market opportunity in the cloud data center and 5G end markets. Marvell has been growing incredibly over the past three years, which I documented here after its recent industry analyst conference. The company has focused and transformed itself into a data infrastructure leader with its storage and compute portfolio, and it also has a strong position in copper networking chips which are used in enterprise and automotive applications. Now, with the Inphi acquisition, Marvell will become a player in optical networking chips used in cloud data centers and wired and wireless carrier networks.
While many people are familiar with Marvell, I think less are familiar with Inphi. Inphi creates the electro-optics chips used by networking equipment makers to connect the massive racks of servers inside cloud data centers, and connect the data centers to one another. The same chips are also used to connect 5G base stations together and to the rest of the network. Consider the cloud giant’s higher speed, lower latency, availability zones, or peering, which all require an ultra high-speed fiber optic connection. Additionally, consider connecting one server rack or fleet to another in the highest speed way possible. Inphi’s chips power those connections.
For wired and wireless carriers, Inphi products connect different parts of the network with fiber optics requiring Inphi’s digital signal processing (DSP) chips, analog support chips, and drivers. As industry data disaggregation increases, which I believe it will, the need increases for fiber optics. What made Inphi so successful was that it cracked the code on higher-order modulation using PAM 4 DSPs for shorter distances and coherent DSPs for longer distances. Since 2018, Inphi has racked up a 40% CAGR and recently had an eye-popping 90% growth for the quarter.
Now, I want to provide the essential news on the deal and my opinion on it.
The news per the Marvell press release and presentation
The following are the deal terms, financial impact, and financing:
- Price: ~$10B, stock, and cash
- Ownership: 83% Marvell, 17% Inphi
- Enterprise Value: approximately $40B
- Accretion: Immediately accretive to Marvell’s revenue growth before synergies. Accretive to non-GAAP EPS growth with synergies
- Proforma P&L: $3.6B revenue, 63.5% GM, $866M (23.81%) opinc, $1.136B EBITDA (31.3%) before synergies
- Cost synergies: $125 million within 18 months of closing the transaction
- Credit facility: $4B in new debt
- Combined TAM: $23B and 12% CAGR
- Deal closure: expected close 2H of 2021
The following are management and governance basics:
- CEO: Matt Murphy
- CFO: Jean Hu
- Board: Inphi CEO Ford Tamer joins Marvell board
Marvell develops silicon for networking (over copper), compute, storage, and security applications for the cloud data center, carrier, and edge markets. Inphi develops silicon for networking (photonics) for the cloud data center and carriers.
This is one of the easier acquisitions to explain for three reasons. First, the two companies serve the same markets, cloud data center, and carriers. Secondly, both are in networking, and Inphi adds the silicon photonics card to Marvell’s arsenal of copper-based networking. This doesn’t mean copper is going away any time soon – it’s not and will have a long life in automotive, the office, the edge, and many enterprise data centers. Fiber optics are required to go longer distances (up to 2km), faster (25-800+ Gbps), and lower power. Third, Marvell built a very strong position in 5G base station market following its acquisition of Cavium in 2018 and Inphi looks to add the same strategic position in cloud data centers. These are two of the fastest growing markets in the industry.
With Inphi solutions inside a switch or router, customers can connect edge and core switches and routers over long distances without the need for a transport box or the internet. And as I have written about previously, I believe in the future, silicon photonics will also be used to connect compute functions versus copper insider servers, which today is done over PCIe and copper.
The two companies also share many cloud, carrier, OEMs, and module customers, so you could imagine a scenario where customers look at Marvell as a one-stop-shop for networking. Marvell is a market share leader in compute for the wireless carrier base stations, and Inphi is the interconnect between the fronthaul and backhaul. In the cloud data center, Inphi is a leader in and between the data center with its photonics interconnects. Marvell then adds its SmartNIC, or now referred to as a DPU and storage networking. In the future, I can see synergy in a combined custom and ASIC opportunity leveraging both Marvell and Inphi. I wrote about Marvell’s ASIC as a service business here.
Unlike some of the larger chip companies, Marvell has a solid competence for M&A, as evidenced by the results of the Aquantia, QLogic, Avera Semi, and Cavium acquisitions. No company is perfect, Marvell included, but many things have been done right when it comes to acquisitions and integrations. I give Marvell CEO Matt Murphy credit for making this happen and his ability to keep the right people on-board and do the right things with them. I expect Marvell to continue this capability with its Inphi acquisition.
Marvell’s move makes sense to me from a technological and customer value proposition sense. Inphi adds that longer range, higher speed, lower power, and higher margin photonics networking capability to Marvell’s copper-based networking. If nothing else, this strengthens Marvell’s “move, store, process and secure” story, no doubt, and makes it even more important to cloud giants and carriers. Both companies are on the upswing, and Inphi is in hyper-growth mode and headed to where the networking puck is going. I have many conversations with cloud hyperscalers and carriers, and the notion of not having everything connected via photonics is laughable. Inphi isn’t cheap, is being acquired at a premium, but as I said before, it is where networking is going, and I believe Marvell will be a much stronger company for it.