In short: Fails the value-add test: "In general, I don't like EMS companies, and you can look at the value add any company has by looking at the gross margins. Celestica, you can see what the gross margins are, get an idea what the value add is." The host notes margins went from ~2% to ~10% — "good expansion but still relatively low." Marvell preferred.
Celestica is an electronics manufacturing services (EMS) company — it assembles hardware, including AI servers, to other companies' designs. Niles doesn't like that business model because it adds little unique value, and gross margin shows it: Celestica keeps far less of each sales dollar than a designer like Marvell. Its margins have improved (from about 2% to about 10% by the host's numbers), but they are still low, so if AI spending slows there is little cushion.
40:54Kind of a great example there. Our next question comes from Mark in Duncan, BC. He's got thoughts on Marvell and Celestica. These are AI spending beneficiaries. What do you do with them? — In general, I don't like EMS companies, and you can look at the value add any company has by looking at the gross margins.
In short: Sole supplier of the Google TPU server + ~50-60% of the cloud-Ethernet-switch market + liquid-cooling lead; bought ~3 yrs ago at ~8× earnings — "decommoditized" critical infrastructure.
Celestica builds the physical guts of AI data centers. For decades it was a low-margin "contract manufacturer" (it assembles hardware for others) — a commodity business. Sacerdote bought it ~3 years ago at about eight times earnings after spotting it was the sole supplier of Google's TPU server and held ~50-60% of the market for cloud "Ethernet switches" (the gear that wires servers together).
This is his "decommoditization" idea: AI machines run so hot they need liquid cooling and cost $200-300k each (vs ~$5k for an old server), so if one breaks the whole system goes down. That turns Celestica's parts into critical infrastructure — like a critical part on a plane that never gets swapped out — giving it real pricing power and durable advantages where there used to be none. ASP = average selling price.
51:05Nvidia for three or four generations in advance we we had this with Celestica Celestica was a contract manufacturer and this has been a disaster industry since 1999. It went all offshore to China. It was commodity, but they hung on and they they kind of kept Celestica's heritage was IBM supercomputing and they kept all that talent and skill.
In short: Cited as one of "the best ways to invest in TPUs" (a TPU hardware supply-chain play) — "I owned one of those." [Transcript "Elastica"; mapped to Celestica.]
Celestica builds the hardware (servers, racks, electronics) that goes into data centers. Baker cites it as one of "the best ways to invest in TPUs" — that is, a supply-chain way to benefit from the build-out of Google's custom AI chips without owning the chip designer itself — and says "I owned one of those."
(Note: the auto-transcript renders the name as "Elastica," which is almost certainly Celestica given the TPU-hardware context. The paired "Momentum" name he mentions is unresolved and not included as a ticker.)
17:15And if somebody is wildly bullish on TPUs today, let's go look at their 13F and let's see if they owned Momentum or Elastica, which were the best ways to invest in TPUs. I owned one of those. And so I do feel like I have some credibility to say this, but I do think Google for a lot of reasons made very conservative design choices with the TPU V8.
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.