Optical Names Fell 10%, Can This Week Fix It?

Nomura: laser chip shipment volume is forecasted to grow from 480 million units in 2026 to 966 million units in 2027, and reach over 1.5 billion units by 2028. CW laser experiences the largest expansion, growing by 402 million units (from 171 million in 2026F to 573 million in 2028F).

The last few days were absolutely brutal.

Citi: According to LightCounting and Frost & Sullivan, the 800G segment achieved an exceptional CAGR of 156.3% from 2021 to 2025 and is forecast to grow at 13.9% from 2025 to 2030E. The 1.6T segment — representing the next generation — is forecast to grow at a CAGR of 98.9% from 2025 to 2030E, with mass adoption forecast to begin in 2026.

$ServiceNow(NOW.US) is on track for its worst day ever, down 24% after earnings

Source: StockMarket.News

Initial notes on $Marvell Tech(MRVL.US)'s rumoured $Lumentum(LITE.US) acquisition:

BLOOMBERG: MARVELL $Marvell Tech(MRVL.US) IN ADVANCED TALKS TO BUY $Lumentum(LITE.US) LUMENTUM HOLDINGS

Source: amit

Mizuho: We believe hyperscalers such as Google and Meta are exploring CPO architectures and could help drive the roadmap, though our call noted volume ramps only in 2028-30E, with current supply chain bottlenecks.

Have heard from a couple experts where InP substrate S/D shortage may come to balance in 2027 with the amount of capacity $AXT(AXTI.US) is ramping.

$Broadcom(AVGO.US)

Post 5 Ep.25

Context: Broadcom has been active in Singapore’s new advanced semiconductor substrate facility. While investors often focus on custom AI accelerators and networking chips, producing these increasingly sophisticated devices requires specialised packaging materials. Broadcom is taking steps to address that bottleneck directly.

My trade: This adds an interesting dimension to my Broadcom thesis. Through its partnership with TOPPAN, the company is strengthening access to the advanced substrates required for its next-generation AI and networking products. For me, this is about more than expanding capacity. It is about reducing supply-chain vulnerabilities that could otherwise limit growth, even when customer demand remains exceptionally strong.

Takeaway: Having the best chip design means little if you cannot manufacture enough of it. In an increasingly capacity-constrained AI industry, securing the ability to deliver may become just as important as securing the customer orders themselves. Sometimes competitive advantage is built long before the finished product reaches the market.

@Captain's Treasure

$NVIDIA(NVDA.US)

Context: NVIDIA caught my attention as AI adoption continues to drive demand for advanced computing and data-centre infrastructure.

My trade: Opened a position to gain exposure to the longer-term AI trend and NVIDIA’s role in providing the chips and technology behind it.

Takeaway: I’ll pay more attention to valuation and expectations around future growth before adding to the position.

[Orders / Holdings card]

$Applied Optoelectronics(AAOI.US)

Post 5 Episode 23

Context: What has strengthened my AAOI thesis is that high-speed optics demand is now translating into actual production pressure. In Q2, revenue reached a record $191.9 million, 800G volumes more than doubled sequentially, and management said demand for 800G and 1.6T products is expected to outpace production capacity through mid-2027.

My trade: That changes how I look at AAOI. The question is increasingly less about whether hyperscalers need faster optical links and more about whether AAOI can execute its manufacturing ramp. The company is targeting roughly 650,000 units of monthly 800G/1.6T capacity by year-end, versus capacity approaching 200,000 currently. Its Q3 revenue guidance of $255–290 million also points to another sizeable step-up if execution follows through.

Takeaway: Growth stories become more compelling when the constraint moves from “Can we find customers?” to “How quickly can we build enough product?” For AAOI, the next chapter may be less about proving the AI-optics opportunity exists — and more about executing fast enough to capture it.

@Captain's Treasure

$Corning(GLW.US)is positioning as a primary supplier of optical infrastructure powering the AI boom. Massive long-term deals have been signed with telecommunications and big tech, with stable recurring cashflow. Currently MA50 is holding on, may change my direction in the short term if the MA50 and subsequent nearby MA200 does not hold

Is It Time to Buy IGV and sell SOXX?

"Buy IGV $iShares Expanded Tech Software Sector ETF(IGV.US)and sell SOXX$iShares Semiconductor ETF(SOXX.US)" trend is a temporary interest rate driven rotation. Trying to dance between hardware and software sectors during a bond market panic is not a good move.

Why?

While it is true that software delivered its strongest single day performance against chipmakers in 25 years, this is not the death of silicon. The extreme cash generation from data center buildouts is still backing semiconductor fundamentals. Software applications can only grow as fast as the hardware chips allow them to compute.

While tech valuations drop when the 10 year Treasury yield hit over 5%, SOXX $iShares Semiconductor ETF(SOXX.US)has already undergone a massive valuation reset down to 20x forward earnings. It is no longer dangerously overpriced.

The Bottom Line

Don't get caught in Wall Street's daily musical chairs. Ignore the temporary software vs hardware rotation headlines. Continue to dollar cost average into QQQM $Invesco Nasdaq 100 ETF(QQQM.US) is the way to go for the long term investor.

It is time in the market that counts, not timing the market.

With 10-Year Treasury yields topping 5%, high rates are hitting capital-heavy chipmakers where it hurts—their long cash-flow duration and massive capex debt. But reading $SOXX vs.$IGV purely as a short-term rate reaction misses the bigger picture.

​This isn't just macro noise; it’s Phase 2 of the AI cycle. We’re moving from the "shovel sellers" (hardware) to the "shovel users" (software) who are actually turning AI into balance sheet revenue through hybrid subscription models. When cost of capital is high, asset-light software generating immediate free cash flow will always win over capital-intensive chip foundries.

​Buy IGV, be ‘selective’ with SOXX.

It looks more like a rotation within the AI trade than a clean hardware to software handoff. The 10-year yield recently crossed 5%, raising the discount rate and pressuring long-duration software valuations.

Meanwhile, AI infrastructure demand remains strong, with hyperscaler capex still accelerating and cloud demand absorbing capacity.

The key signal to watch would be whether AI software revenue and margins start accelerating faster than infrastructure spending. If that happens alongside sustained high yields, the hardware to software rotation thesis becomes much stronger.

everything that's goes up must come down $Lumentum(LITE.US)

The surge in chip demand makes total sense—AI agents shifting from cloud workloads to mainstream consumer hardware require massive on-device processing power. It’s more than just hype; it's a structural shift in how CPUs and edge chips are utilized