Fabian, first (ever) of his name. Senior analyst at @FundaAI covering networking hardware. Not investment advice. Not any advice. My own views.
🇱🇧🇷🇴🇬🇷🇩🇪🇺🇸
Joined May 2009
- Tweets20.6K
- Following4K
- Followers6.6K
- Likes114K
Pinned Tweet
Happy to share that I am joining @FundaAI.
For more than a year, the team has been publishing rigorous, differentiated research on photonics, memory, software, and the rest of the AI infrastructure stack. I have spoken to them several times over the past few months and have been impressed by the depth of their expertise, their culture, and their determination to build a respected and trusted research firm. On top of that, they are building an ambitious AI powered research OS that I think will change how the buy side consumes fundamental work.
I will continue running my own newsletter on Substack alongside the FundaAI role, and the two should complement each other well.
A note on prior content: any posts, tweets, or newsletter pieces published before my start date at FundaAI reflect my personal views at the time of writing and were produced entirely independently of FundaAI. They do not represent the views of FundaAI, were not reviewed or endorsed by FundaAI, and should not be interpreted as the firm's research, recommendations, or investment advice. Any positive or negative commentary on specific companies, securities, or individuals in that prior content predates my affiliation with FundaAI.
Going forward, all tweets and posts on my newsletter represent my personal views only and do not constitute investment advice or the views of FundaAI. There will be some restrictions on what I can comment on. FundaAI takes its regulatory obligations seriously and expects the same from its team.
It has been an incredible six months on Substack, life changing to say the least, and I am looking forward to the next leg of this journey.
I still believe this will go down as a terrible decision for Nvidia if it happens.
BREAKING: Nvidia, $NVDA, is preparing to fire Jensen Huang, its founder and CEO, after allegations of a second affair have surfaced, per WSJ.
The board is preparing to meet as soon as today to determine his future.
3
9
2,024
I am seeing a decent number of startups working on "wide and slow" but I am also hearing legacy semi names looking to pivot into that direction.
Another microLED startup -- Staera -- with some interesting per-lane rates (4-8 Gbps), and they want to go scale in too.
- 4 to 8 Gbps per link, with more than 600 links per millimeter of chip edge
- Designed for under 0.2 pJ/bit, against 5 pJ/bit for co-packaged optics
- More than 100x the performance of electrical links across the millimeter-to-meter range
- Reaches beyond 50 mm, against 2 mm for UCIe-A die-to-die links
- Lightwire-based HBM designed to exceed 24 TB/s, against 2.8 TB/s per HBM4 stack today
2
2
775
It is business as usual and even more all hands on deck than ever. Where do these reporters get their stories man?
unavailable
An honorable competitor to Palantir's Maven. It was bound to happen. I look forward to see if/how customers (if they even announce them publicly) will adopt the technology. "Under Europe control", does that mean it is GDPR-compliant?
Introducing Clarity – an AI-native agentic intelligence platform for armed forces.
Developed in Europe, under European control.
helsing.ai/clarity/
5
1,053
I am so looking forward to the @TomorrowXSummit and their elite lineup. I am more than looking forward to see @McConaughey mediating on the sessions.
Alright alright alright!!!!!
$MRVL / $LITE main takeaways from Marvell's latest SEC filings (Nov. 2–10), including Matt Murphy's internal employee email and interviews with CNBC and Fox Business.
1. Stock selloff and investor concerns
Marvell shares have declined ~51% over the past month, broadly in line with the 49% average decline among AI connectivity peers. Lumentum has experienced a similar correction. Despite the selloff, Marvell remains up ~66% YTD.
Marvell CEO Matt Murphy attributes the weakness to concerns over the sustainability of AI infrastructure spending, semiconductor cyclicality and potential restrictions on data center development following the US midterm elections.
He also acknowledged that initial analyst commentary questioned the acquisition's strategic rationale, although conversations with major institutional shareholders in Boston and New York were reportedly more constructive.
2. Why acquire Lumentum now?
Matt Murphy believes optical connectivity is approaching another major architectural transition, moving beyond scale-out into scale-up and scale-in applications. @FundaAI has argued in the past months that both Marvell and Lumentum are the best positioned firms for scale-in applications and the acquisition will help speed up these developments.
He compared the opportunity to Marvell's acquisition of Inphi, whose scale-out optical business reportedly grew approximately 10x from a ~$700M annualized revenue base over five years.
Importantly, he argues that hyperscalers are making architectural decisions today that will determine the next generation of AI infrastructure. Waiting could mean missing critical design cycles.
1/3
3. $550B combined TAM
Marvell estimates its standalone TAM at $400B, compared with Lumentum's $90B.
Management believes the combination creates another $60B in addressable opportunities, bringing the total to approximately $550B.
The incremental opportunity is largely concentrated in scale-up networking, particularly NPO and CPO, where the companies could move from selling individual components to delivering complete optical solutions.
4. NPO/CPO and system-level integration
The strategic objective is to integrate Marvell's semiconductor technology with Lumentum's lasers and optical components.
Rather than selling DSPs, optical components and light sources independently, management envisions integrated modules and solutions that can accelerate hyperscaler qualification, time-to-market and production deployments.
Matt Murphy also noted that both companies were able to cross-check customer demand during diligence because their technologies participate in the same system designs.
5. Lumentum's InP and laser differentiation
Matt Murphy directly challenged the view that Lumentum's laser business is simply benefiting from temporary supply shortages.
He described its InP and laser manufacturing technologies as highly proprietary, requiring specialized processes and expertise that are difficult to replicate.
His argument is that differentiation should persist even after supply constraints normalize.
2/3
1
7
814
👀👀
We believe if Jensen was removed as CEO, Nvidia is worth a lot less (25%-30%) as a company/stock and would change the future direction of the company and create a dramatically different competitive environment around AI GPUs globally especially for China. ~15% chance he removed (as of now) in our view
6
1,915
Santec management does not see any signs of a slowdown and raised revenue guidance by 35% and OP by 50%.
As @TheStalwart mentioned in one of his videos here, we are all susceptible to the Dunning-Kruger effect, and the events of the past week (especially the past 24 hours on X) have shown just how easily it can play out in real time.
5
712
"Nvidia Corp.’s board is launching an investigation into whether Chief Executive Officer Jensen Huang was targeted by a foreign espionage plot after he admitted to employees that he was unfaithful to his wife." bloomberg.com/news/articles/…
2
869
@FundaAI's SpaceX orbital compute update was published just before this video dropped, where @elonmusk reported a initial order for 6,000 space-specific racks. We laid out a bull case of getting 1GW of compute into space by end 2028.
fundaai.substack.com/p/deeps…
Elon Musk just revealed that SpaceX will begin launching StarMind SI computers into orbit in Q3 2027, with major deployment ramps in 2028 and 2029. The first systems are being co-designed with NVIDIA around Vera Rubin NVL72. Space-based supercomputing is becoming a reality
9
904
