Elon Musk Bought Into POET Technologies' Thesis — With Light

Musk Just Paid Up for the Optical Thesis. POET Is the Public Version of the Same Bet.

When the world's most aggressive capital allocator buys an optical-interconnect company, it says something about where the AI buildout is headed next — and about the one small-cap already standing in that lane.

On June 25, 2026, the Federal Trade Commission granted early termination of its antitrust review for Elon Musk's acquisition of Mesh Optical Technologies — a Los Angeles startup founded by former SpaceX $SPCX engineers that had been out of stealth for barely four months. The filing lists Musk personally as the acquiring party under transaction number 20261601. Financial terms were not disclosed. Bloomberg first reported the deal.

Take a step back from the Musk headline and look at what he actually bought, and a clearer signal emerges: the next front in the AI infrastructure war will not be fought over raw compute. It will be fought over how fast, how efficiently, and how cheaply data can move between the chips — with light. That is a thesis $POET Technologies (NASDAQ: POET) has been executing against for years.

What Musk Actually Bought

Mesh's core product is the Alpha C1, an optical transceiver that converts electrical signals to light at 1.6 terabits per second, supporting both 800G and 1.6T applications. Its pitch rests on two claims: dramatically lower power — the company says roughly a third of the power of rival modules — and, critically, manufacturability. Mesh builds its optical engine using flip-chip die bonding, a high-volume semiconductor packaging technique, rather than the slow, manual assembly that has historically bottlenecked optical hardware.

The strategic logic is obvious once you map Musk's other assets. xAI runs Colossus, one of the largest GPU clusters on the planet. SpaceX has signed multi-year compute agreements reportedly worth tens of billions of dollars and is floating a plan — internally codenamed Starmind — for space-based AI compute across a future satellite constellation. Every one of those ambitions runs into the same wall: AI training is increasingly bottlenecked not by GPU count but by data movement. Optical interconnect at 1.6 Tbps is how you knock that wall down.

Why This Is the Whole Ballgame

Strip the deal down and it is a validation event. One of the most capital-rich, technically ruthless operators in the world just decided that owning optical-interconnect technology outright was worth moving on within months of the company's public debut. He did not build a GPU. He did not buy more compute. He bought the plumbing that lets compute scale.

That is precisely the argument POET has made to the market: that photonic integration — moving data with light instead of copper — is the enabling layer for the entire AI buildout, and that the winners will be decided less by peak optical performance in a lab than by who can manufacture at volume, at low power, at low cost. Musk did not validate a company. He validated a category. And POET is one of the few public, small-cap ways to express a bet on that category directly.

The Manufacturability Moat — and the Interesting Contrast

Here is where the comparison gets sharper. Both companies are chasing the same 800G-to-1.6T optical-engine market, and both anchor their story on manufacturability rather than raw specs. But they attack the packaging problem from different levels of integration.

Mesh's differentiator is flip-chip die bonding — a precise, high-volume way to assemble the optical engine at the component level. POET's approach starts a level deeper. Its patented POET Optical Interposer integrates electronic and photonic devices onto a single chip using wafer-level semiconductor manufacturing. In plain terms: rather than bonding finished parts together efficiently, POET builds the integration into the wafer itself. The company's stated payoff is the same list Mesh is chasing — lower cost, lower power, smaller size, and a design that scales to high production volumes. Much of this was just discussed in the company’s annual meeting.

The takeaway is not that one approach obviously wins. It is that the smartest money in technology just confirmed the axis of competition POET has been betting on: whoever industrializes optics — turns it from artisanal assembly into semiconductor-grade manufacturing — controls the next phase of AI infrastructure.

Where POET Sits Right Now

POET is not a concept stock waiting for a market to arrive; the market just arrived on its doorstep. At its June 26, 2026 Annual General Meeting, management laid out where the business stands. According to the company, POET now has more than ten active customer engagements that combined are expected to exceed $100 million in future annual revenue — including a recently announced agreement with Lumilens, a production order for 800G optical engines, and a significant NRE agreement for external light source applications from a new customer.

On the manufacturing side, POET reported that its optical-engine production ramp remains on schedule to begin in the second half of 2026, with capacity targeted at up to one million units per month by the end of 2027, supported by roughly $50 million in planned capital-equipment spending. The company says it raised $830 million in equity over the trailing twelve months and could raise up to $661 million more if outstanding warrants are exercised, and it plans to grow its 115-person workforce by roughly 50 in the coming months.

In other words: while a Musk-backed startup is just now scaling its first product, $POET is already booking customer engagements, ordering the equipment, and telling the market it intends to ship optical engines in volume this year.

The Honest Counterpoint

Validation cuts both ways. A category worth Musk's attention is a category that will attract Musk-sized capital, and a well-funded, vertically integrated competitor with a captive demand sink across xAI and SpaceX is a genuine competitive factor, not a footnote. Mesh does not need to sell to the open market to matter; internal demand alone could make it formidable.

And POET's own numbers are, at this stage, forward-looking. The customer-engagement revenue, the capacity target, and the production timeline are company projections and stated plans, not booked results — the ramp still has to execute, and the broader optical-module market already includes entrenched incumbents alongside new entrants. Readers should weigh the thesis against POET's execution risk, its history of equity issuance, and the competitive intensity that a deal like this one signals is only increasing.

Bottom Line

Connect the dots and the picture is hard to unsee. The most aggressive capital allocator of this era just spent to own optical-interconnect technology, on the logic that light — not copper, not more raw silicon — is the constraint that decides how far AI can scale. That is the exact thesis POET has been building manufacturing capacity and customer engagements around. Musk validated the war. POET is already on the battlefield.