Atomera: Betting on Quantum-Engineered Silicon at 2nm

Atomera's core intellectual property is Mears Silicon Technology (MST) — a quantum-engineered thin film of reengineered silicon, typically 100 to 300 angstroms thick (roughly 20 to 60 atomic unit cells). MST works by inserting ultra-thin layers of oxygen into silicon wafers at precise intervals, creating a superlattice structure that enhances the electrical properties of semiconductor devices. The technology addresses a fundamental physics problem: as transistors shrink below 3nm, dopant atoms (like phosphorus) tend to migrate out of their intended locations during manufacturing, degrading performance and yield. MST acts as a diffusion barrier, keeping dopants where they belong. Atomera's business model is IP licensing — they don't manufacture chips or wafers at scale. The revenue model has three components: (1) Non-recurring engineering (NRE) fees for custom wafer deliveries and development work, (2) MSTcad software licensing for simulation tools, and (3) the long-term prize — production royalty licenses where chipmakers pay per-wafer fees to use MST in volume manufacturing. Currently, virtually all revenue comes from NRE and MSTcad, which generated a combined $65,000 in FY2025. The company has also explored a 'wafer-based' go-to-market approach where MST is deposited onto blank wafers before the customer begins their manufacturing process, reducing integration complexity and lowering the adoption barrier. Atomera targets multiple semiconductor markets: GAA logic (for AI chips at 2nm and below), DRAM memory, RF silicon-on-insulator (RF SOI) for 5G, power semiconductors (trench FETs for data centers), and gallium nitride on silicon (GaN-on-Si) for next-gen RF and power devices. The GAA opportunity is the most time-sensitive and potentially transformative — TSMC and Samsung are both ramping 2nm production in 2026-2027, and if MST becomes part of their process flow, the royalty stream could be substantial. However, semiconductor adoption cycles are notoriously long, and no production license has been signed to date.

FY2025 total revenue: $65,000. Yes, sixty-five thousand dollars for a full fiscal year. Q4 2025 revenue was $50,000, up 117% YoY from $23,000 but below the $102,000 consensus estimate. GAAP net loss: $20.2 million ($0.65/share) vs. $18.4 million ($0.68/share) in FY2024. Non-GAAP net loss: $16.1 million ($0.52/share) vs. $15.4 million ($0.57/share) in FY2024. The loss widened primarily due to a $1.1 million increase in stock-based compensation from new performance stock units (PSUs) and $794,000 in higher R&D spending as Atomera transitioned to new device fabrication vendors after ending its relationship with TSI Semiconductors. Non-GAAP operating expenses were $15.9 million in FY2025 — $10.2 million in R&D (up from $9.4 million) and $4.8 million in G&A (down from $5.1 million). Cash and short-term investments ended FY2025 at $19.2 million, down from $26.7 million a year earlier — a 25.5% decline. The company raised approximately $7.6 million through its at-the-market (ATM) equity offering in 2025 and an additional ~$3.2 million post-year-end. Total liabilities are a modest $2.71 million with no long-term debt. At the current quarterly burn rate of ~$3.4 million (including the post-year-end ATM proceeds bringing cash to roughly $22 million), Atomera has approximately 6-7 quarters of runway — taking the company into late 2027 before requiring additional capital. Q1 2026 revenue guidance: $50,000-$100,000 from MST wafer shipments. The company does not provide guidance beyond the current quarter. Insider activity is notable: CEO Scott Bibaud, CFO Francis Laurencio, and CTO Robert Mears collectively sold 17,406 shares over the past six months at an average price around $2.39, with zero insider purchases. While these are small amounts (likely RSU-related tax sales), the optics are poor for a stock trading on faith in management's vision.

Atomera operates in a unique niche — engineered silicon materials for advanced semiconductor manufacturing — where direct public comparisons are limited. The closest competitive dynamics involve: (1) Internal R&D at major fabs — TSMC, Samsung, and Intel all have massive materials science teams working on diffusion barriers and channel engineering. If they solve the GAA dopant migration problem internally, Atomera's value proposition diminishes. (2) Silicon arsenic — the incumbent diffusion-blocking material, which Bibaud described as 'expensive and dangerous' and inferior to MST in testing. (3) Alternative materials companies — Diamond Foundry (synthetic diamond substrates), Porotech (porous GaN), and GaN Systems (acquired by Infineon) operate in adjacent semiconductor materials spaces but don't directly compete with MST's silicon enhancement approach. (4) In the broader IP licensing model, Atomera competes for fab attention alongside other materials and process innovation providers. Atomera's moat, to the extent one exists, is its patent portfolio — the company holds over 300 patents and patent applications covering MST's composition, deposition methods, and device applications. The technology has been validated through peer-reviewed publications and presentations at major semiconductor conferences. The partnership with a 'large equipment OEM' (unnamed) for GAA applications provides a potential channel for broader adoption. However, the moat is ultimately only as strong as the manufacturing proof points — semiconductor fabs won't adopt a new material based on patents alone. They need extensive reliability data, yield impact studies, and integration validation, which is exactly what Atomera is working through now.

1. GAA customer engagement expansion (H1 2026): Bibaud stated a goal of engaging all four major GAA customers (TSMC, Samsung, Intel, Rapidus) following recent 2nm silicon results. Any announced joint development agreement would be a significant de-risking event. 2. DRAM customer data readout (March 2026): Final results from a DRAM customer are expected 'in about a month' from the February earnings call. Positive results could trigger a JDA and production license discussion. 3. GaN commercial customer progress (2026): Atomera's first commercial GaN-on-silicon customer has begun running wafers and can both grow GaN wafers and manufacture devices — potentially moving faster than Atomera's academic collaborations. 4. PowerAmerica funding decision (H1 2026): Atomera's GaN-on-silicon concept paper advanced to the proposal stage, which could provide external development funding and industry validation. 5. First production license agreement (2026-2027): The ultimate catalyst. Any signed production license — even at modest initial volumes — would fundamentally change the investment narrative from 'science project' to 'commercializing technology company' and likely trigger a significant re-rating.

• Cash runway risk: With $19.2M in cash (plus ~$3.2M raised post-year-end) and a ~$13.5M annual burn rate, Atomera needs to raise additional capital by late 2027. Dilution is inevitable — the only question is the share price at which it occurs.
• Zero commercial traction after 10+ years: Atomera was founded in 2001. Despite 24 years of development, annual revenue has never exceeded $500K. The semiconductor industry's adoption cycle is long, but this timeline tests even patient investors' faith.
• Customer concentration and dependence on major fabs: Atomera's entire thesis depends on TSMC, Samsung, Intel, or another major fab adopting MST into volume production. These companies have enormous internal R&D budgets and no obligation to license external technology.
• Insider selling with zero insider buying: All three named executives sold shares in the past six months while none purchased. Combined with major hedge fund exits (Millennium -100%, Morgan Stanley -76%, D.E. Shaw -66% in Q3 2025), smart money has been heading for the exits.
• Technology obsolescence risk: The semiconductor industry moves fast. A competing diffusion-blocking approach, or a change in transistor architecture that makes MST unnecessary, could render the technology irrelevant before it reaches commercial scale.