
The industry powering AI, explained by @vikramskr and @austinsemis. Weekly podcast and daily takes. Sign up at https://nitter.cf/t.co/wSYNiNJYqL and subscribe on YT
Joined January 2026
- Tweets1.8K
- Following4
- Followers10.6K
- Likes608
Pinned Tweet
🎙️ NEW EPISODE: Why Marvell Wants Lumentum: Vertical Integration in AI Optics With Cignal AI
Recorded the morning after the $MRVL $LITE rumor broke. Our thoughts (live with no prep!)
Andrew Schmitt and Scott Wilkinson of market research firm Cignal AI join Austin and Vik to break down Marvell's potential acquisition of Lumentum. They explain how Marvell could run the "Broadcom playbook" by bundling DSPs and optics, even at the cost of competing with its own customers.
- The driver is vertical integration, not securing laser capacity. Marvell wants to control the supply chain from chips to optics to co-design better products for AI
- Marvell's DSP customers would have to compete with a vertically integrated Marvell-Lumentum, a dynamic that has already pushed some players out of coherent
- Lumentum's Optical Circuit Switches (OCS) and high-power lasers for CPO and NPO matter as much as its current-gen transceivers
- After a 4x expansion of its Indium Phosphide fab, Lumentum can supply Marvell AND keep selling lasers to the open market
- A bidding war is possible. Could the leak be intentional, to attract suitors?
This episode is presented by @CrusoeAI. Crusoe's Serverless Fine-Tuning trains an open model like Llama, DeepSeek, or GLM on your own data, and you pay per million tokens, not per GPU hour. About $2 buys roughly 5M training tokens on smaller models. Get $5 in credit to start: semidoped.com/crusoe-x
Chapters:
0:00 Intro: Marvell & Lumentum
7:38 The Vertical Integration Driver
9:17 The Channel Conflict Problem
14:58 Transceiver Value Chain
17:38 Why Would Lumentum Sell?
21:23 Lumentum's InP Capacity
24:52 The Cost of Waiting
27:00 Manufacturing Scale Is the Moat
31:30 How AI Changed Component Value
33:43 CPO: The Integration Endgame
36:50 The 'Broadcom Playbook'
38:58 Prediction: A Bidding War
Get more of Austin and Vik daily, free!
Sign up: daily.semidoped.com/
Connect with Vik and Austin:
Vik's Paid Substack: viksnewsletter.com
Austin's Paid Substack: chipstrat.com
@austinsemis @vikramskr @aschmitt @ScottWilkinsonQ
Intel Plunges 23% as PC Weakness and Supply Snags Crush Outlook
Austin: Bummed to continue to hear Intel unable to keep up with server CPU demand, this has been going on for many quarters. And the other side of the products business seemingly has too much supply into a down client market (due to memory prices, which we’ve known for a long time that memory would hurt client).
If you enjoy our pod, please check out our sponsor @CrusoeAI using the link below. Sign up quick or log in with Google and get $5 in credit!
When you claim that $5 in credit, this show keeps going! 🙏
semidoped.com/crusoe-x
🎙️ NEW EPISODE: Why Marvell Wants Lumentum: Vertical Integration in AI Optics With Cignal AI
Recorded the morning after the $MRVL $LITE rumor broke. Our thoughts (live with no prep!)
Andrew Schmitt and Scott Wilkinson of market research firm Cignal AI join Austin and Vik to break down Marvell's potential acquisition of Lumentum. They explain how Marvell could run the "Broadcom playbook" by bundling DSPs and optics, even at the cost of competing with its own customers.
- The driver is vertical integration, not securing laser capacity. Marvell wants to control the supply chain from chips to optics to co-design better products for AI
- Marvell's DSP customers would have to compete with a vertically integrated Marvell-Lumentum, a dynamic that has already pushed some players out of coherent
- Lumentum's Optical Circuit Switches (OCS) and high-power lasers for CPO and NPO matter as much as its current-gen transceivers
- After a 4x expansion of its Indium Phosphide fab, Lumentum can supply Marvell AND keep selling lasers to the open market
- A bidding war is possible. Could the leak be intentional, to attract suitors?
This episode is presented by @CrusoeAI. Crusoe's Serverless Fine-Tuning trains an open model like Llama, DeepSeek, or GLM on your own data, and you pay per million tokens, not per GPU hour. About $2 buys roughly 5M training tokens on smaller models. Get $5 in credit to start: semidoped.com/crusoe-x
Chapters:
0:00 Intro: Marvell & Lumentum
7:38 The Vertical Integration Driver
9:17 The Channel Conflict Problem
14:58 Transceiver Value Chain
17:38 Why Would Lumentum Sell?
21:23 Lumentum's InP Capacity
24:52 The Cost of Waiting
27:00 Manufacturing Scale Is the Moat
31:30 How AI Changed Component Value
33:43 CPO: The Integration Endgame
36:50 The 'Broadcom Playbook'
38:58 Prediction: A Bidding War
Get more of Austin and Vik daily, free!
Sign up: daily.semidoped.com/
Connect with Vik and Austin:
Vik's Paid Substack: viksnewsletter.com
Austin's Paid Substack: chipstrat.com
@austinsemis @vikramskr @aschmitt @ScottWilkinsonQ
Nvidia Weighs Credit Guarantees to Back Nscale’s Anthropic Compute Deal
Vik: If NeoCloud (like NScale) isn't getting enough customers, what benefit is there in NVIDIA stepping in, purchasing compute, and renting it back? It's just more bandaid measures.
Austin: I disagree, I think broadly neoclouds are helpful and necessary but access to capital is hard, and like a parent cosigning a mortgage with their first-time homebuying child, Nvidia can help neos get past bank headaches.
Daily Update - October 30th, 2026
- Nvidia weighs credit guarantees for Nscale Anthropic compute deal
- Samsung’s record profit runs on memory
- Intel plunges 23% on PC weakness
- TSMC nears deal to buy two AUO fabs
- Murata and Samsung Electro-Mechanics race to supply AI server capacitors
- Hanmi Semiconductor hits 50% operating margin on HBM TC bonder surge
daily.semidoped.com/p/daily-…
We had on the good folks from @CignalAI on Semi Doped to help us think through the rumors of Marvell acquiring Lumentum.
Episode should be out by the weekend.
Stay tuned!
How do you solve the package IO bottleneck for AI scaling?
Intel Foundry is exploring complementary optical interconnects, using the top surface of the package to send light in and out for a massive increase in bandwidth density.
Intel Foundry's Ravi Mahajan explains.










