$AXTI has clear fundamental exposure to rising InP demand, but in the context of this specific news, I would view it as a company facing policy risk rather than as a clean beneficiary.
AXT is listed in the United States, but its wafer production is concentrated in China, and exports of its InP substrates already require Chinese government approval. If U.S. scrutiny expands beyond Chinese transceiver vendors to include the origin of upstream materials, $AXTI could face additional customer qualification requirements, tariffs, or supply chain risks.
The potential benefit to $TSEM is more indirect, but its exposure is relatively cleaner from a policy perspective. If this shift leads customers such as $COHR and $MRVL to expand silicon photonics production outside the Chinese supply chain, $TSEM could benefit over the longer term.
That would be a much bigger escalation. It would affect not just Chinese transceiver vendors, but potentially the entire global optical supply chain. Let’s hope it doesn’t come to that.
Informative article. Thanks for posting. I question your comment about POET being an “indirect or long term benefactor (paraphrase)”. I assume you state this because POET is not currently shipping in volume, or you question the timeframe for qualification by major Hyperscalers. I’d argue that POET has focused on scaling production for some time now, and they are building toward a 1 million unit per month capacity - greater than AAOI is currently achieving. I also believe that qualification has already been completed within at least one major hyperscaler, and that orders will be announced at a point where POET has demonstrated an ability to deliver - likely a trigger point being the shipment of the 30,000 units this quarter. I just think they are closer than you assume, and I believe they are a direct beneficiary of this Innolight “lock out”. Otherwise, I love your articles and greatly appreciate your work.
Thank you for sharing your perspective. I respect your view.
I have also started watching POET closely again recently. In particular, I am keeping an eye on the rumor that the Marvell deal could be revived following the CFO’s retirement. As you may know, I did a very deep dive on this company in the past.
As you said, I sincerely hope this turns out to be good news for POET as well.
I’m glad to hear you’re starting to watch POET closely again.
Personally, I don’t believe they need to revive the Marvell deal in order to become a very successful company. Not only have they already established an ecosystem of partners to bring solutions to the AI market (including, I believe, a channel already selected by a major hyperscaler that has already qualified POET’s solution), but there are other markets for POET’s solutions - Telecomm, LiDAr, etc. The inflection point will be the first shipments and follow on orders - Marvell or no Marvell. IMHO
Great note, thanks. What is your view on impact to $AXTI or $TSEM?
$AXTI has clear fundamental exposure to rising InP demand, but in the context of this specific news, I would view it as a company facing policy risk rather than as a clean beneficiary.
AXT is listed in the United States, but its wafer production is concentrated in China, and exports of its InP substrates already require Chinese government approval. If U.S. scrutiny expands beyond Chinese transceiver vendors to include the origin of upstream materials, $AXTI could face additional customer qualification requirements, tariffs, or supply chain risks.
The potential benefit to $TSEM is more indirect, but its exposure is relatively cleaner from a policy perspective. If this shift leads customers such as $COHR and $MRVL to expand silicon photonics production outside the Chinese supply chain, $TSEM could benefit over the longer term.
What if chinas answer to trumps import restrictions are export controls on indium phosphide 💀?
That would be a much bigger escalation. It would affect not just Chinese transceiver vendors, but potentially the entire global optical supply chain. Let’s hope it doesn’t come to that.
AEHR announced a third agreement/hyperscaler? customer today.
👍👍👍👍👍
Informative article. Thanks for posting. I question your comment about POET being an “indirect or long term benefactor (paraphrase)”. I assume you state this because POET is not currently shipping in volume, or you question the timeframe for qualification by major Hyperscalers. I’d argue that POET has focused on scaling production for some time now, and they are building toward a 1 million unit per month capacity - greater than AAOI is currently achieving. I also believe that qualification has already been completed within at least one major hyperscaler, and that orders will be announced at a point where POET has demonstrated an ability to deliver - likely a trigger point being the shipment of the 30,000 units this quarter. I just think they are closer than you assume, and I believe they are a direct beneficiary of this Innolight “lock out”. Otherwise, I love your articles and greatly appreciate your work.
Thank you for sharing your perspective. I respect your view.
I have also started watching POET closely again recently. In particular, I am keeping an eye on the rumor that the Marvell deal could be revived following the CFO’s retirement. As you may know, I did a very deep dive on this company in the past.
As you said, I sincerely hope this turns out to be good news for POET as well.
I’m glad to hear you’re starting to watch POET closely again.
Personally, I don’t believe they need to revive the Marvell deal in order to become a very successful company. Not only have they already established an ecosystem of partners to bring solutions to the AI market (including, I believe, a channel already selected by a major hyperscaler that has already qualified POET’s solution), but there are other markets for POET’s solutions - Telecomm, LiDAr, etc. The inflection point will be the first shipments and follow on orders - Marvell or no Marvell. IMHO