First, let me correct a premise: the company currently rumored to be in talks to acquire Lumentum (LITE) is Marvell (MRVL), not Coherent (COHR). So what I want to ask about is the COHR side: is there any possibility that Coherent itself participates in the bidding, or conversely acquires Lumentum? If so, please give specific numbers on the motivation and financial feasibility based on the two companies' financial reports, production capacity, and equity structure: (1) the gap between Coherent and Lumentum in InP production capacity, FY2026 revenue, and gross margin structure, and where Coherent's PhotonLink platform stands; (2) how Lumentum's production capacity being sold out through early 2029 and its share count diluted to 1.011 yi shares in the June 2026 quarter affect the acquisition consideration and dilution; (3) after Marvell's three-step integration (Celestial AI, XConn, Lumentum) connects chips plus light sources, what would Coherent gain and lose as the only independent large-scale light source supplier; (4) what impact Nvidia's March 2026 investment of 20 yi US dollars each in Lumentum and Coherent has on the two sides' equity structure, board seats, and the feasibility of any transaction. Please give specific numbers and sources; do not give buy or sell recommendations.

First, let me correct a premise: the company currently rumored to be in talks to acquire Lumentum (LITE) is Marvell (MRVL), not Coherent (COHR). So what I want to ask about is the COHR side: is there any possibility that Coherent itself participates in the bidding, or conversely acquires Lumentum? If so, please give specific numbers on the motivation and financial feasibility based on the two companies' financial reports, production capacity, and equity structure: (1) the gap between Coherent and Lumentum in InP production capacity, FY2026 revenue, and gross margin structure, and where Coherent's PhotonLink platform stands; (2) how Lumentum's production capacity being sold out through early 2029 and its share count diluted to 1.011 yi shares in the June 2026 quarter affect the acquisition consideration and dilution; (3) after Marvell's three-step integration (Celestial AI, XConn, Lumentum) connects chips plus light sources, what would Coherent gain and lose as the only independent large-scale light source supplier; (4) what impact Nvidia's March 2026 investment of 20 yi US dollars each in Lumentum and Coherent has on the two sides' equity structure, board seats, and the feasibility of any transaction. Please give specific numbers and sources; do not give buy or sell recommendations.

Summary of Key Points

This matter has not yet been officially announced, but three 8-K filings have already laid the cards on the table: Nvidia in March took three stakes of 20 yi USD each in Lumentum, Coherent, and Marvell, turning the light-source layer from "bought on the market" into "strategically allocated."

Lumentum is the fastest runner this round (FY26 revenue 30.14 yi USD, +83.2%, GAAP gross margin 41.7%), while Coherent's scale is 2.4 times its own but its growth rate is only a quarter of Lumentum's, so the question is not who gets bought out, but how much bargaining power the remaining independent light-source vendor still has.

To his question of "will Coherent turn around and snatch it": sufficient motive, insufficient money — available liquidity 26.5 yi versus 830 yi consideration, a 31-fold gap. On the two points of sold out through 2029 and PhotonLink, I cannot find them; the article marks them item by item.

I. First, Correct a Premise

The opening of the question says "the one in talks to acquire Lumentum is Coherent, not Marvell" — this half-sentence is wrong. Our piece from beginning to end wrote Marvell; Coherent appears 6 times in the full text, and every time it is in the position of "the only remaining independent large light-source vendor," not the acquirer. If the questioner saw "Coherent acquires Lumentum" elsewhere, that item is wrong; do not reason further along with it.

According to Bloomberg's 2026-10-28 report, the one in talks is Marvell, at approximately 830 yi USD. As of 2026-10-29, Lumentum's latest 8-K stops at 2026-08-11 (earnings report), with no 425, no S-4, no DEFM14A — this deal has not yet entered statutory disclosure proceedings. So all judgments below about "post-acquisition" are built on the premise of "if it lands," not on established fact.

II. First Question: The Two Companies' Revenue and Gross Margin Structure, InP Capacity, PhotonLink

First look at structure; the numbers are all taken from each company's most recent complete fiscal year (both companies' fiscal years end at the end of June).

Lumentum FY2026 (ended 2026-06-27): revenue 30.14 yi USD, year-on-year +83.2%; of which Components 20.06 yi, +79.7%, Systems 10.08 yi, +90.7%. GAAP gross margin 41.7% (prior year 28.0%, up 13.7 points in one year), non-GAAP 46.0%; GAAP operating margin 17.4%, non-GAAP 29.8%. By the single quarter it is even steeper: FY26Q4 revenue 10.1 yi, GAAP gross margin 47.4%, non-GAAP 50.4%, GAAP operating margin 27.8%. Guidance for the next quarter is revenue 12.25–12.75 yi, non-GAAP operating margin 39.5%–40.5%.

Coherent FY2026 (ended 2026-06-30): revenue 71.18 yi USD, year-on-year +22.5%; GAAP gross margin 37.5% (prior year 35.2%, up 2.3 points), non-GAAP 39.4%; GAAP operating margin 12.6%, net profit 8.05 yi. Single quarter FY26Q4 revenue 20.46 yi, year-on-year +33.8%, GAAP gross margin 38.5%, non-GAAP 40.2%.

Putting these numbers together, three things are clear:

First, the scale differs by 2.4 times, and the growth rate differs by 3.7 times. Coherent does 71 yi USD a year, Lumentum does 30 yi; but Lumentum is year-on-year +83.2%, Coherent +22.5%. This is not a distinction of "big versus small," but of "who is eating the increment of this round of AI optical interconnect."

Second, gross margin differs by 4.2 points (GAAP 41.7% versus 37.5%), and the gap is widening rather than narrowing — Lumentum raised 13.7 points in one year, Coherent raised 2.3 points. The source of this difference is not product mix, but capacity utilization: Lumentum's GAAP gross margin over these four quarters is 34.0%→36.1%→44.2%→47.4%; a production line already at full load can still climb like this, showing it is picking customers and picking prices; Coherent spreads across a broad product line, so it rises slowly.

Third, "Systems" accounts for one third of Lumentum's revenue, and its growth rate is even faster than Components (+90.7% versus +79.7%). This point is often overlooked: it does not only sell lasers, it is also moving toward modules and subsystems. So what Marvell buys in buying it is not only light-source capacity, but also a segment of module capability that has already been built up.

Regarding the comparison of InP capacity, I want to tell the truth: neither company discloses wafer capacity.

In Coherent's FY26 10-K, "InP" appears 43 times, but all are qualitative descriptions (which product lines use indium phosphide, which fabs are doing it), with not a single wafer, not a single production line number. Lumentum likewise does not disclose. So any claim that "Company A's InP capacity is several times Company B's," as long as no source is given, is an estimate.

There are two hard metrics that can serve as substitutes: first, the slope of gross margin climb (the set of numbers above), which reflects "whether the production lines can still squeeze out more"; second, capital expenditure. Both companies are expanding production in this round, but the pace of expansion differs greatly — Lumentum's gross margin climbed up without major expansion, indicating its existing production lines have been squeezed to the limit; Coherent is raising while expanding, so the slope is naturally flat.

PhotonLink: cannot be found in primary documents.

In the full text of Coherent's FY26 10-K, PhotonLink appears 0 times; in the FY26Q4 earnings release (8-K Exhibit 99.1) it is also 0 times. I have not seen this platform name in the company's statutory disclosures. If the questioner saw it in a press conference, investor day, or sell-side report, please provide the source, and I will verify again against that document. Before then, I cannot make up an answer for it about "what position it is in."

III. Second question: how "sold out through early 2029" and "1.011 yi shares" affect consideration and dilution

First, the phrase "capacity sold out through early 2029."

I searched Lumentum's FY26Q4 earnings release, the 2026-08-11 8-K, and the full text of the FY26 10-K (keywords 2029, sold out, fully booked, capacity), and did not find this expression. Where 2029 appears in the 10-K is "2029 Notes" — that is its convertible bonds maturing in 2029, unrelated to capacity.

This phrase very likely comes from the Q&A session of the earnings call, or from sell-side channel checks. It is not false, but it is not a primary disclosure, and I cannot use it as evidence. To really confirm, the way is to listen to the FY26Q4 earnings call recording, or wait for the next earnings call (FY27Q1, about November) for management to say it themselves.

Now the share count, for this I found exact numbers.

Lumentum's common stock on 2026-06-27 was 88.6 million shares, versus 69.8 million shares in the same period last year — an increase of 27% in one year. The reason for the increase is not a new issuance, but debt-to-equity conversion: the company converted three convertible bonds — 2026, 2028, 2029 — into common stock, recording 78 yi USD of debt extinguishment loss (non-cash) for the full year. At the same time, preferred stock was 2.9 million shares, the stake Nvidia bought in March, convertible into common stock at 1:1.

So the "fully converted" share count is 88.6 + 2.9 ≈ 91.5 million shares. The 1.011 yi shares the questioner mentioned is about 960 wan shares more than this basis, with the difference probably coming from options and restricted stock (the diluted share count in the 10-K notes would be needed to verify precisely, which I did not retrieve this time).

The per-share prices calculated under the three bases differ by 12%, and this is the actual meaning of "dilution":

Based on 830 yi USD total consideration, divided by 88.6 million shares it is about 937 USD per share; divided by 91.5 million shares (including the Nvidia preferred stock conversion) it is about 907 USD; divided by 1.011 yi shares it is about 821 USD.

For the acquirer, this is not an accounting detail, it is real money: for every additional 1,000 wan diluted shares, the 830 yi total price requires 8 yi USD more to be spread out. So when negotiating such a deal, the first thing is to fix "fully diluted share count" — how options are handled, at what price preferred stock converts, how much debt-to-equity conversion remains unconverted.

Nvidia's 2,876,415 preferred shares convert into common stock at 695.31 USD/share, and the conversion price is locked. If the transaction price is above 900 USD, it earns this spread along with it, and its cost is 20 yi USD in cash; if the deal falls through, it remains a preferred shareholder of Lumentum. This position is "no loss either way" for it, and for the acquired company's original shareholders it is "one more person sharing."

IV. Third question: after Marvell connects "chip + light source," what does Coherent gain and what does it lose

First, let's be clear about what Marvell is actually fighting for this year. On 2026-02-02 it spent 3.534 billion USD to buy Celestial AI (optical interconnect fabric), in the first half of the year it spent 469 million USD to buy XConn (CXL switching), on 10-22 it raised its revolving credit facility to 3 billion USD, and six days later came reports it was in talks for Lumentum. The three steps connect into one line: DSP (which it already had) + optical interconnect fabric + switching + light source. The light source is the last piece, and the only piece where "if you can't buy it, you can only queue."

What Coherent gains is three things.

First, it becomes "the only independent large light-source vendor that belongs neither to the module makers nor to the DSP makers." Cloud providers dual-source; on Marvell's side it is self-supplied; the one remaining vendor can only be Coherent. This position does not require it to be very good—only to still be there.

Second, Nvidia's March investment was not just money. Item 7.01 of the 8-K states clearly: the two sides also have a collaboration, and Nvidia gains access to 5 additional Coherent product families related to co-packaged optics (CPO). That is, it entered Nvidia's CPO supply-chain list—not "one of the suppliers," but "the one it co-develops with." This matters more than 2 billion USD.

Third, the breadth of its product line has become an advantage. Lumentum is strong in light sources and modules; Coherent has everything from materials (silicon carbide, indium phosphide) to devices to systems, and is more complete on "one-stop." During integration periods, customers fear switching suppliers most; breadth is the insurance.

What it loses is one thing: neutrality.

After Nvidia took a stake, it went from "a supplier anyone can buy from" to "a company with a strategic shareholder." Before other cloud providers place large orders with it, they will ask two questions: Will my next-generation product specs reach Nvidia first? When capacity is tight, where do I rank? These two questions did not exist before; now they do. The result is not lost orders, but every large order will carry an extra "dual-source commitment" clause—which is precisely why, after Marvell acquires Lumentum, cloud providers will rely on Coherent more but also guard against it more.

Working backward from FY26 net profit of 805 million and diluted earnings per share of 4.12 USD, Coherent's diluted share count is about 195 million shares, and Nvidia's 7,788,161 shares account for roughly 4%. 4% is not enough to control, but enough to have a say in key negotiations.

In one sentence summarizing this question: what Coherent gains is "position," and what it loses is "neutrality." Position can be converted into orders; neutrality can only be converted into clauses.

V. The fourth question: the impact of Nvidia's three 2 billion USD investments on equity, the board, and deal feasibility

All three are private placements, not secondary-market purchases—this point is important.

Lumentum (2026-03-02, 8-K Item 3.02): issued 2,876,415 shares of Class A convertible preferred stock at 695.31 USD per share, totaling 2 billion USD in cash, under the Section 4(a)(2) private-placement exemption. These preferred shares can convert into common stock 1:1.

Coherent (2026-03-02, 8-K Item 3.02): issued 7,788,161 shares of common stock at 256.80 USD per share, totaling 2 billion USD in cash. Item 7.01 of the same 8-K disclosed the CPO collaboration with Nvidia and access to 5 product families. The company was explicit about the money's use: for R&D, future capacity expansion, and operational capabilities, in support of expanding its U.S. domestic manufacturing.

Marvell (2026-03-31, 8-K Item 3.02): issued 2,000,000 shares of Class A convertible preferred stock, totaling 2 billion USD in cash, convertible into at most 21,778,000 shares of common stock (implying a conversion price of about 91.8 USD per share).

Board seats: not in the primary documents.

The 8-Ks of Lumentum and Marvell both disclose the certificate-of-designation terms for the preferred stock, one of which states plainly: the preferred stockholders have no voting rights on the election of directors (on other matters they vote together with common stock on an as-converted basis). Coherent's 8-K has only Items 3.02, 7.01, and 9.01—no seat arrangement, no standstill, no voting agreement.

So the claim that "Nvidia obtained board seats through these three investments" currently has no basis. What it obtained is equity and commercial collaboration, not governance rights.

But the impact on deal feasibility is greater than seats.

NVIDIA is simultaneously a preferred shareholder of Lumentum, a common shareholder of Coherent, and a preferred shareholder of Marvell. If any party wants to acquire Lumentum, it faces the same question: how would this company's largest strategic customer and shareholder view the deal?

It does not need veto power to make this work. It can decline to renew supply agreements, can give next-generation CPO specifications to another company, and can adjust the order of capacity allocation. None of this requires a meeting or a vote. So the question "will NVIDIA agree" is really "how many suppliers does NVIDIA want to keep at the light-source layer, and who does it want to be the consolidator."

From its action of placing three bets simultaneously, the answer is: it does not bet on a single company; it bets on "having me in all three." This means that consolidation by any one of them will not change its position—this is the most convenient design and also the hardest design to circumvent.

The three real boundaries of the transaction are still the same three: Chinese regulatory approval (the probability of conditions being attached is higher than prohibition), whether cloud providers insist on dual sourcing, and the physical pace of light-source capacity. NVIDIA's three bets do not change these three, but they make the second one harder—because the other source in the dual-source setup is now also its shareholder.

V-2. The question he asked: if Coherent instead acquires Lumentum, the motive is sufficient but the money is not

This question was in the earliest two submissions; the later one (the one holding the key) deleted it. Since it is being handled as an independent question, I will answer it separately.

The motive is sufficient. What Coherent fears most right now is Marvell combining DSP and light sources into one and becoming the only company able to supply everything in one stop. If it could seize Lumentum, it would in turn hold that position itself—this is not just defense; it is its only opportunity this year to change the landscape. It also has NVIDIA's list of 5 CPO product families, so in theory this makes sense.

But the money hurdle cannot be overcome, and it is not a small shortfall—it is short by an order of magnitude.

Coherent's available liquidity as of 2026-06-30 was: cash 11.62 yi, short-term investments 8.25 yi, and 6.64 yi remaining under its revolving credit facility, totaling about 26.5 yi; at the same time it had 32.22 yi in total debt (including 9.9 yi of 5% senior notes, 28 yi of revolving facility borrowings, and 14 yi of German factory construction loans).

Lumentum, based on the 10-28 close of 913.95 US dollars and 91.5 million shares, has a market value of about 836 yi US dollars. In other words, the acquiree is more than twice the size of the acquirer—Coherent's own market value is about 468 yi (195.4 million shares × 239.59 US dollars), only 56% of Lumentum's.

Both paths are blocked:

All cash: it would need to borrow 830 yi. Its current total debt is 32.22 yi, equivalent to increasing it to 26 times in one go. No banking syndicate would approve it, and the rating would immediately fall to junk.

All stock: issuing new shares at 239.59 US dollars would require issuing about 3.46 yi shares; adding the existing 1.95 yi shares, original shareholders would be diluted to 36%. This is not an acquisition; this is handing itself over—and Marvell's 20 yi US dollars of preferred stock is still watching from the sidelines.

The reverse direction (Lumentum acquiring Coherent) looks feasible—836 yi versus 468 yi—but is equally unrealistic: Lumentum has just converted three convertible bonds into stock, has book cash of 20.44 yi plus short-term investments of 6.95 yi, and is being dogged by acquisition rumors; at this point it will not undertake an M&A deal more than twice its own size.

There is another number worth his attention: 830 yi US dollars versus 836 yi market value means the premium is almost zero. This shows that the market has already priced this deal as "close to completion." In this situation, once a third-party bidder appears, the price increase will not be 5%, but double digits—and Coherent cannot afford it.

So the conclusion is: Coherent's rational choice is not to compete for it, but to sell itself at a good price, or to hold the position of "the only independent large light-source supplier" and wait for pricing power when capacity is tighter. What it has truly obtained this year—NVIDIA's 4% stake and the CPO list—is not for attack, but for not being circumvented at the negotiating table.

VI. What I did not find out, listed item by item

• Lumentum "capacity sold out until early 2029" — no such statement found in the full-text search of the FY26Q4 earnings release, 8-K, and 10-K; it may have come from the earnings call and was not accepted.
• The share count basis of 1.011 yi shares — what I found is 88.6 million common shares + 2.9 million preferred shares (approximately 91.5 million fully converted), a difference of about 960 wan shares suspected to be options/RSUs, requiring precise verification against the diluted share count in the 10-K notes.
• Coherent's PhotonLink platform — appears 0 times in both the 10-K and the Q4 earnings release; its position cannot be determined.
• The specific figures for the two companies' InP wafer capacity — neither discloses them, and any multiple claim lacks first-hand basis.
• Whether Nvidia signed standstill agreements, rights of first refusal, or priority supply terms with the three companies — none of the three 8-Ks discloses this; only equity and (in Coherent's case) the CPO collaboration.

Sources

• Lumentum 8-K, 2026-03-02, Item 3.02 / 5.03, accession 0001193125-26-085412 (Nvidia $2 billion preferred shares)
• Coherent 8-K, 2026-03-02, Item 3.02 / 7.01, accession 0001193125-26-084366 (Nvidia $2 billion common shares + CPO collaboration)
• Marvell 8-K, 2026-03-31, Item 3.02 / 5.03, accession 0001193125-26-134462 (Nvidia $2 billion preferred shares)
• Lumentum FY2026Q4 earnings release, 2026-08-11, 8-K Exhibit 99.1 (revenue 30.14 yi, GAAP gross margin 41.7%, Q4 47.4%, FY27Q1 guidance)
• Lumentum FY2026 10-K, 2026-08-17, accession 0001628280-26-057358 (88.6 million common shares, 2.9 million preferred shares, 78 yi debt conversion loss)
• Coherent FY2026Q4 earnings release, 2026-08-12, 8-K Exhibit 99.1 (revenue 71.18 yi, GAAP gross margin 37.5%, Q4 38.5%)
• Coherent FY2026 10-K, 2026-08-14, accession 0000820318-26-000020 (InP-related statements, no capacity figures)
• Marvell FY2027Q2 earnings release, 2026-08 (data center revenue +46%, non-GAAP gross margin 58.9%)
• Marvell 10-Q, 2026-08-28 (Celestial AI 35.34 yi USD, XConn 4.69 yi USD)
• Bloomberg 2026-10-28 report (Marvell in talks for Lumentum, approximately 830 yi USD, not officially announced)
• Lumentum 8-K, 2026-08-11, accession 0001628280-26-055726 (most recent periodic disclosure, no 425 / S-4)
• Coherent 8-K, 2026-08-31, accession 0001193125-26-375462 (other disclosures during the quarter)