Wall Street loves a good infrastructure play, and few themes command more attention in 2025 than artificial intelligence (AI). While most investors focus on chipmakers and cloud providers, a less obvious bottleneck is emerging: Data centers can't move information fast enough. Traditional electrical interconnects hit physical limits around power consumption and heat generation at the speeds AI workloads demand.

Poet Technologies (POET -4.23%) is building silicon photonics solutions designed to solve this problem by using light instead of electricity to transmit data between chips. On Oct. 7, 2025, the company announced its largest-ever financing -- a $75 million private placement with a single institutional investor.

Is Poet Technologies stock a buy on this capital infusion? Below, I'll consider the company's technology, partnerships, and financial reality to find out.

A semiconductor with the letters AI glowing on it.

Image source: Getty Images.

Light moves faster than electricity

Poet's core technology is the Optical Interposer, a platform that integrates electronic and photonic components onto a single chip using wafer-level manufacturing. The company designs optical engines for 800G and 1.6T data transmission speeds -- the backbone of next-generation AI clusters and hyperscale data centers.

The technology addresses a real problem. AI accelerators and high-performance computing systems generate massive amounts of data that need to move between processors in real time. Electrical interconnects consume significant power and generate heat, limiting how fast data can flow. Optical interconnects use light to transmit information, offering higher bandwidth, lower power consumption, and less heat generation.

Poet has secured partnerships with major players, validating its approach. In May 2024, Foxconn Interconnect Technology selected Poet's optical engines for its 800G and 1.6T optical transceiver modules.

On Sept. 30, 2025, Poet and Semtech launched 1.6T optical receivers for AI networks. The same week, Poet announced a collaboration with Sivers Semiconductors on external light sources for co-packaged optics targeting the AI market.

The $150 million war chest

On Oct. 7, 2025, Poet raised just over $75 million from a single institutional investor, lifting total cash on hand above $150 million, with no significant debt. The funds are earmarked for research and development, acquisitions, the light-source business, and working capital.

This capital base gives Poet the ability to push its technology toward commercial scale. With Foxconn planning volume production of 800G and 1.6T modules in the second half of 2025, Poet targets meaningful revenue as hyperscalers expand AI infrastructure in 2026 and beyond. The timeline for broad adoption, however, remains uncertain.

The revenue reality check

Poet reported just $268,000 in revenue during the second quarter of 2025, up from none a year earlier. Net loss for the quarter reached $17.3 million, with trailing-12-month revenue totaling only $468,000. Commercial production has yet to begin. Current sales come from engineering services and product samples, not volume manufacturing.

At a market capitalization of about $736 million, Poet's stock reflects expectations of future adoption rather than current fundamentals. Driving this point home, Poet's shares trade at over 52 times 2026 projected sales.

What's the bottom line? Investors are speculating on partnerships, technology validation, and the long-term demand for optical interconnects in AI infrastructure.

A speculative infrastructure play

Poet sits at the intersection of massive AI-driven demand and an unproven business model. The company has technology validation, top-tier partners, and $150 million in cash but has yet to show meaningful commercial revenue. That combination makes the stock highly speculative.

For aggressive investors, Poet provides exposure to silicon photonics as a potential backbone of AI infrastructure. For cautious investors, the prudent approach is to wait until scaled production and recurring revenue are visible before assigning value to the story.