@talethdabm89 The research firm got their glass core substrate timelines off.
H2 2026 should be production start (SKC Absolics), not 2028. And TAM figures seems off toward 2040, so thought it was better not to cite it.
Even though it’s not full value chain like $LPK and GCS in specific
- $AAOI at $12B
- $SIVE at $2B
- Foci at $2.8B
- Shunsin at $2B
Usually the best risk/reward to me currently. Lot of my answers before like $AXTI already 10x’d, so different lineup this time.
$AAOI due to absurd H1 2027 revenue projections from capacity ramp, doing everything from laser fab to assembly in America.
$471M/month… that’s in 2027, the TAM increases exponentially in 2028.
$SIVE is also ramping absurdly high, 77% revenue pipeline growth of the entire company’s history to ~$799M
Primarily from photonics… in a single quarter. And they’re projecting 60% gross margins off that.
Foci - $NVDA / $TSM primarily FAU supplier and bottleneck for COUPE. Genuinely not sure how this is $2.8B.
BOM share for their passive components + FAU are massive in 2028. Just a bit early H1 2026.
Shunsin - Legit you see Foxconn get CPO/photonics related orders over and over for $NVDA and others.
Just nobody knows the packaging/testing gets done by Shunsin.
A lot of contracts are also under Shunsin’s subsidiary too.. so markets/algorithms don’t know what’s coming imo.
Runner up is $XFAB, they’ll probably be central to EU CHIPS act 2 for silicon photonics at ~$1.5B MC.
And of course SiC/GaN foundries should go brr with 800vdc push by Nvidia.
Especially if they’re the only high volume one in United States per Dpt. Of Commerce.
And it’s such a low price/book ratio so you’re kinda getting the company upside for free, while US Gov/EU Gov subsidize their capex.
做 AI Agent Demo 不难,难的是把它真正部署到云上:环境、评估、CI、数据接入,每一步都能劝退人。
Google 开源的 agents-cli,思路是把这些工程化流程做成 AI 编程助手可以调用的技能包。
GitHub:https://t.co/q654jCGX6G
它可以接入 Claude Code、Gemini CLI、Codex 等工具,让编程 Agent 帮你完成脚手架搭建、模型评估和云端部署。
也可以脱离 AI 助手单独运行,用命令行完成数据接入和持续集成相关操作。
如果你不是只想做一个 Agent Demo,而是想把它往真实项目里推进,这个项目值得关注。