Rising Rates Are Repricing the AI Photonics Trade

The chip selloff this week didn’t start in a chip. It started in the Treasury market.

The 10-year yield is back near 4.6%. The 30-year cracked 5%. Import prices just printed +4.2% year-over-year, the hottest since 2022. CPI is sitting near a three-year high. And the bond market — which spent the last cycle assuming the Fed would ride to the rescue — is now pricing in roughly a 40% chance of a rate hike by December, not a cut. New Fed Chair Kevin Warsh hasn’t even run his first press conference yet. That happens June 17, and nobody knows which Warsh shows up: the “regime change” hawk from his speeches, or the steady hand markets are praying for.

That’s the cause. Here’s the effect — and why it lands hardest on exactly the names building the AI data center.

Why rates hit growth stocks in the teeth

This is the part retail keeps relearning the hard way. A stock is worth the present value of its future cash flows. The discount rate you use to drag those future dollars back to today is anchored to the risk-free rate — the Treasury yield. When that yield climbs, every dollar a company earns five, ten, fifteen years out is suddenly worth less in today’s terms.

Now ask which companies have the most of their value sitting way out in the future. Not Coca-Cola. It’s the high-multiple growth names whose valuations are built almost entirely on cash flows they haven’t earned yet. That’s the entire AI-semiconductor complex. So when yields back up, these are the stocks that get repriced the fastest and the furthest. It’s not a referendum on whether the AI buildout is real. It’s arithmetic.

The longer the duration of a company’s cash flows, the more violent the move. And that’s the key to understanding why the optical and photonics names — POET, Lightwave Logic, Lumentum, Coherent — aren’t just falling with the group. They’re falling more.

The duration ladder inside the AI hardware stack

Think of these names as rungs on a ladder, sorted by how far out their cash flows sit.

Coherent (COHR) and Lumentum (LITE) are the closest to “now.” They’re real businesses with real revenue — Coherent did roughly $5.8 billion last year, Lumentum about $1.65 billion — selling transceivers and optical components into AI networking today. They get hit when rates rise, but they have current earnings to cushion the blow. Lumentum’s CEO has said capacity is sold out through 2028. That’s a near-term cash flow story, and near-term cash flows are the least sensitive to the discount rate.

POET Technologies (POET) sits further out on the ladder. The optical interconnect thesis is compelling and the engagement is accelerating, but the bulk of the value the market is paying for lives in future design wins and future ramp — not last quarter’s revenue line. Push the discount rate up, and a name like this reprices harder than a Lumentum, because more of its worth is parked in years that just got mathematically cheaper.

Lightwave Logic (LWLG) is the far end of the ladder. Pre-revenue, pure-thesis, electro-optic polymer materials that could be transformative if they hit commercial scale. Nearly 100% of the valuation is future cash flow. That makes it the single most rate-sensitive profile in the group. When the risk-free rate spikes, a pre-revenue name doesn’t get a discount — it gets a markdown.

Same macro shock. Wildly different magnitude of pain, depending on where your cash flows live on the calendar.

But duration isn’t destiny — capital is

Here’s where the ladder gets a twist, and it’s the part that separates POET from the rest of the speculative end of the group.

Being a long-duration story makes you price-sensitive to rates. But what determines whether you actually survive a long, high-rate stretch is the balance sheet. And on that score, POET and Lightwave Logic are not in the same weight class.

POET went into Q1 2026 already sitting on roughly $429 million in cash and short-term investments against an operating cash burn of only about $8.8 million a quarter. Then, in mid-May, it added another $400 million gross in a single registered direct offering — and the terms of that raise are the whole story. POET priced the deal at $21.00 a unit, a premium to the $20.57 prior close, to a single institutional investor, with three-year warrants struck at $26.15. Read that back slowly: a premium, in one block, to one buyer, with warrants priced 25% above the deal. That is not how a capital-starved company raises money. Desperate names raise at a discount through dilutive at-the-market drips because nobody will write a big check. POET got the big check, at a premium, on its own terms.

Stack that on the existing balance sheet and POET is carrying something on the order of $800 million in capital against a single-digit-millions quarterly burn. That’s not months of runway, it’s years of it — enough to fund a roughly ten-fold manufacturing capacity expansion through 2027 without ever putting a gun to its own head in a bad tape. Better still, the cash is partly self-funding the burn: POET booked about $4 million in other income last quarter, much of it interest earned on that pile. The war chest throws off enough yield to offset a meaningful chunk of the cash going out the door — and that was before the new $400 million started earning.

Lightwave Logic doesn’t have that luxury. Pre-revenue with a cash position in the low tens of millions and a structural reliance on tapping the equity market to keep the lights on, LWLG has to raise capital while its stock is getting repriced by rising rates. That’s the trap: the exact macro environment that craters the share price is the environment in which a thin-balance-sheet name is forced to sell more shares to fund operations — diluting holders at the worst possible moment, at whatever price the market will bear. Rate stress and dilution risk feed each other in a doom loop.

This is the difference between weathering a drawdown and being forced to act inside one. POET can sit on its hands through six, eight, ten ugly months of high yields and simply keep building, letting the macro turn on its own schedule. A capital-light name has a clock running, and the market knows it — which is part of why the discount gets applied so aggressively.

POET’s cash position gives it far more freedom than LWLG to absorb the peaks and valleys of the market. In a game of runs, capital carries you to the finish line. The thesis can be identical; the staying power is not. When the discount rate is the enemy, the company that just raised $400 million at a premium — and never has to be a forced seller of its own equity — has already won half the battle.

Why this group is doubly exposed

The optical names carry a second beta on top of the rates problem, and it compounds the first.

These are AI-infrastructure derivatives. They don’t sell the GPU — they sell the plumbing that lets racks of GPUs talk to each other without melting on power and bandwidth. That makes them a leveraged bet on hyperscaler capex. So they ride two narratives at once: “is the AI buildout still on?” and “what’s the discount rate on that buildout?” When both turn negative in the same week — capex-bubble chatter from the BofA strategist crowd, plus a yield spike — the smaller, less liquid component names get the worst of both.

And liquidity matters here. NVDA can absorb a wave of selling. A POET or an LWLG cannot. Thinner float means the same dollar of selling pressure moves the price further. On red days, that’s why these names gap down past where the fundamentals say they should. On green days, it’s also why they rip — the door is narrow in both directions.

The other half of the squeeze: oil and the Iran wildcard

There’s a feedback loop running underneath all of this. The reason inflation re-accelerated is partly energy — the Middle East conflict drove crude toward $104 and import prices with it. Over the long weekend, hopes of a US-Iran deal and a Strait of Hormuz reopening knocked oil back toward $97, which should cool inflation and ease the yield pressure.

But here’s the trap: the same “peace trade” optimism that’s good for inflation is rotating money out of crowded AI names and into financials, energy, and autos. So even the good macro news isn’t clean for this group right now. Lower oil helps the discount-rate story but hurts the relative-flows story. And reports of fresh US military strikes over the weekend put the whole peace narrative back in question. The optical names are caught in the crossfire of a geopolitical story they have nothing to do with.

What actually breaks the logjam

For this group to catch a sustained bid rather than a dead-cat bounce, the macro has to cooperate. The cleanest catalyst is the inflation print cooling enough to take the December hike off the table and pull the 10-year back down toward — and through — 4.3%. That single move does more for POET and LWLG than any press release they could issue, because it directly lowers the discount rate crushing their out-year cash flows.

The second catalyst is Warsh on June 17. If he signals patience and stability instead of regime change, long-duration growth gets relief. If he leans hawkish, the repricing has another leg.

And the third is company-specific proof: a design win, a revenue ramp, a named hyperscaler customer. Hard fundamental news pulls cash flows forward on the calendar, which is the only company-controllable way to fight a rising discount rate. It’s the antidote to duration risk.

The takeaway

Don’t confuse the mechanism with the verdict. Rising yields repricing the AI-component complex is not the market saying the optical buildout is fake. It’s the market saying future dollars got more expensive this week, and the names with the most future and the least present got repriced the hardest. That’s $COHR and $LITE bending, $POET buckling more, and $LWLG taking the full force — exactly in order of how far their cash flows sit from today.

If you believe the photonics thesis, rate-driven drawdowns are the part of the cycle where the thesis gets tested against your conviction, not against the company’s fundamentals. But conviction isn’t enough on its own — the company has to be able to outlast the drawdown without diluting you to death. That’s where POET’s war chest separates it from the pack: a balance sheet that was already deep before it pulled in another $400 million at a premium, leaving it with the kind of capital that turns a multi-quarter rate siege into a non-event. The bond market is setting the price of patience right now, and patience is something you can only afford if you’ve got the capital to wait. The question for each of these names is whether you’ve got the duration to match theirs — and whether their balance sheet has the duration to match the cycle.