# The Laser Chokepoint Sold at a 12% Discount: What the Spread Says *URL:* https://millionaur.com/millionaur-2026-11-03-daily-analysis/ *Updated:* 2026-11-03T07:59:08+00:00 On Monday someone paid about 115 billion USD for the laser bottleneck I have been writing about for months. The market then sold the buyer, the target and Korea’s chip makers within 24 hours. Here is how I read it, and what the portfolio did. ## The deal Marvell (MRVL) agreed to buy Lumentum (LITE) in an all-stock deal: 4.5 Marvell shares for each Lumentum share, about $1,132 per share when announced, closing in the first half of 2028. Lumentum makes the high-power lasers inside AI optical links, and says it is sold out into 2029. The useful number is not 115 billion. It is the gap. After Marvell fell again, the deal is worth about $767 per Lumentum share, while Lumentum trades near $686. That is a spread of almost 12% on a deal that needs 18 months and many regulators. The market is openly pricing a chance that it never closes. My read: Lumentum is now Marvell paper plus a deal-break risk, so I am not buying it. The interesting part is everyone else. Every chip designer that competes with Marvell now has a reason to qualify a second, independent laser supplier. Coherent (COHR) and Applied Optoelectronics (AAOI) report in the next days. That is where I look for evidence, not in share-price moves. ## Memory: the buyer said it out loud Apple (AAPL) beat on revenue but guided its holiday-quarter gross margin to 45-46%, below expectations. Its CEO called memory pricing “unprecedented” and said costs keep rising beyond December. That hurts companies where memory is a cost. It supports companies where memory is the revenue. The biggest consumer hardware buyer is paying more, not cancelling. Korea still sold off on Tuesday (KOSPI -3.6%, SK hynix -5.7%), on US AI-funding nerves before the elections, not on memory prices. ## Funding: open, but paid in shares Oracle (ORCL) filed for $20 billion of new money ($10 billion in bonds, $10 billion in new shares), five days after calling $20 billion the final amount it needed. 17 companies issued high-grade bonds before the vote, so credit is open. But buyers now fund AI with equity too. I keep preferring suppliers paid from customers’ cash flow over buyers who issue paper. ## What the portfolio did - Oil fell about 7% on Iran diplomacy. Three energy stops filled at their exact levels, all in profit, and the LNG position closed with about +10%. Energy went from 7.6% to 5.7% of the portfolio. - On the dip I rebuilt memory: SK hynix back in, more DRAM, and a reset on Micron (MU) at a lower price. - I added trailing stops on Amazon after its recovery. - Cash is about 17%, plus inflation bonds, Treasuries and staples. Tech and AI are only about 20% of the portfolio, so there is room to add when the evidence comes. ## What I am watching - Tonight: AMD results and the US midterm elections. - This week: Lumentum, Applied Optoelectronics and Coherent results, then Friday’s jobs report. - The Marvell-Lumentum spread: if it shrinks below about 5%, the market believes the deal will close. - Oil: a failed Iran deal would bring the risk premium back fast. ## Risks, stated openly - Positioning: October saw record hedge-fund selling. It can continue even when the news is good. - Funding: more equity issuance by AI buyers could slow spending across the chain. - Events: elections and earnings can gap prices through stop levels. ## The takeaway True facts, falling prices. That is what a positioning market looks like. My job this week is not to guess the results. It is to have the triggers written down, so the right move is obvious after them. *Portfolio snapshot: 3 November 2026, 07:44 UTC (US prices = Monday 2 November closes; Seoul = Tuesday close).* Sources: [Marvell press release](https://investor.marvell.com/news-events/press-releases/detail/1040/marvell-and-lumentum-to-combine-creating-a-category-leader-in-ai-connectivity), [9to5Mac on Apple’s margin guide](https://9to5mac.com/2026/11/02/apple-says-memory-costs-and-siri-ai-will-weigh-on-holiday-quarter-profit-margins/), [The Next Web on Oracle’s offerings](https://thenextweb.com/news/oracle-10b-bond-sale-share-offering), [Seoul Economic Daily on Korea](https://en.sedaily.com/finance/2026/11/03/samsung-sk-hynix-slide-on-us-ai-selloff-kospi-tests-5700). **This is not financial advice.** Past performance does not guarantee future results, and copy trading carries a real risk of loss. Please invest only what you can afford to lose. --- Human page: https://millionaur.com/millionaur-2026-11-03-daily-analysis/?vwao_human=1