🐦 推特SerenityV1 · 实盘观望把握 3/10
北京时间·🕒 发布 04-28 01:31:37时区推断·⟳ 抓取 06-03 19:00:13回填·⚡ 决策 06-17 06:15:52·132.4 天前
KOL推演Marvell可能将POET踢出光引擎供应链并转向Sivers/MACOM/Lumentum
📄 原文Serenity↗ 原文
I've always maintained $POET can get designed out and likely would in a few years for $MRVL. Especially as hyperscalers…
不做 POET.US
仅为KOL基于传闻的推演，无一手信源证实MRVL剔除POET；小盘股做空易遭逼空，未证实传闻不碰
不做 MRVL.US
供应链微调对MRVL整体营收和估值影响极小，无edge
观望 MTSI.US
KOL推演MRVL多源采购利好MACOM，但无实锤订单，等产业链交叉验证
展开详情(成交明细 / 研究档案 / 原文全文 / 决策记录链)
研究档案
这条决策没有研究档案。
原文全文
展开原文 · 1,377 字
I've always maintained $POET can get designed out and likely would in a few years for $MRVL.
Especially as hyperscalers tend to vertically integrate upstream, I just didn't expect it so soon.
For me to be bullish, $POET would have needed to multi-source to multiple hyperscaler supply chains rather than just Marvell.
However, this goes this show:
There's a reason why all the Laser companies are worth tens of billions.
Laser design and fab is much more difficult than companies like $POET buying the lasers to package them.
If $MRVL were completely reliant on $POET, they wouldn't break off the engagement despite the NDA, so it does feel like an excuse. Poet was just the fastest time to market given it's already designed into Celestial, but serves more as a Gen-1 bridge.
Regardless all the hyperscalers require a light source (which is much much more difficult to vertically integrate):
And after $NVDA went and bought out allocation for $LITE, $COHR...
There's not many laser suppliers left aside from $SIVE.
There's a lot of nuances markets might miss, but packaging layers are different than laser layers.
It's highly likely now $MRVL just buys lasers directly from Sivers (which is even more bullish) given they already match Celestial specifications.
And also highly likely Marvell multi-sources with $MTSI and $LITE (if there's any capacity left).
Especially as hyperscalers tend to vertically integrate upstream, I just didn't expect it so soon.
For me to be bullish, $POET would have needed to multi-source to multiple hyperscaler supply chains rather than just Marvell.
However, this goes this show:
There's a reason why all the Laser companies are worth tens of billions.
Laser design and fab is much more difficult than companies like $POET buying the lasers to package them.
If $MRVL were completely reliant on $POET, they wouldn't break off the engagement despite the NDA, so it does feel like an excuse. Poet was just the fastest time to market given it's already designed into Celestial, but serves more as a Gen-1 bridge.
Regardless all the hyperscalers require a light source (which is much much more difficult to vertically integrate):
And after $NVDA went and bought out allocation for $LITE, $COHR...
There's not many laser suppliers left aside from $SIVE.
There's a lot of nuances markets might miss, but packaging layers are different than laser layers.
It's highly likely now $MRVL just buys lasers directly from Sivers (which is even more bullish) given they already match Celestial specifications.
And also highly likely Marvell multi-sources with $MTSI and $LITE (if there's any capacity left).
决策记录链 · 只追加
每次领取产生一条记录,重新决策追加一条并以 supersedes 指向被替代的记录(ADR 0014)。
这条决策早于决策账本(2026-10-08 起逐条追加),没有追加记录。