ATOM — Q2 2026 Financial Report Analysis

Atomera's second quarter of 2026 looked like most of its quarters: almost no revenue, a widening loss, and a bank balance kept healthy by selling new shares. Revenue was $158,000, all of it engineering-services income from delivering test wafers, not license fees. The net loss widened 28% to $6.3 million because operating expenses rose 33%, driven mostly by bonus accruals, stock-based pay, patent legal costs and a larger sales team. Loss per share stayed flat at $0.17 only because there were 27% more shares outstanding to spread the loss across. A February share sale doubled cash and short-term investments to $38.4 million, which at the first half's burn rate covers a little over two years of operations.

At a glance

• $158,000 of revenue: still trivial next to $6.9 million of quarterly operating costs. It came from delivering MST-coated wafers to customers for evaluation, not from a license or royalty, so it says little about whether the business model is starting to work.
• $6.3 million net loss, up 28%: costs grew faster than anything else. General and administrative expense rose 55% and sales and marketing tripled.
• $38.4 million in cash and short-term investments: up from $19.2 million at the end of 2025 after a $23.6 million (net) share sale in February. That funds roughly 2.25 years at the current cash burn of about $8.5 million per half-year.

What Atomera does, and why revenue is so small

Atomera does not make chips. It licenses a process called Mears Silicon Technology (MST): a very thin layer of re-engineered silicon added to a wafer during manufacturing, which the company says makes transistors faster, less leaky and more power-efficient, using equipment chip factories already own. The business plan is to earn licensing revenue in stages. A customer pays an R&D license fee when it installs MST on a tool in its factory, then a larger high-volume manufacturing (HVM) license fee once it qualifies MST for production, and then ongoing royalties on every MST-enabled chip it sells.

The 10-Q shows the company is still before that first big step for most customers. Revenue to date has come only from engineering services, integration licenses, R&D licenses under joint development agreements, a license agreement with STMicroelectronics, and MSTcad simulation-software licenses. The filing's own risk language says the company has "only recognized minimal engineering services and licensing revenues." Without an HVM license and royalties, the income statement is mostly a cost statement.

Q2 2026 results

Percentage changes in revenue, net income and EPS are marked n/m (not meaningful). Revenue grew from zero, and a loss compared with a smaller loss produces a percentage that is technically correct but tells you nothing useful. Operating margin, meaning operating profit as a share of revenue, is left out for the same reason: a $6.8 million operating loss on $158,000 of revenue works out to roughly -4,300%, which does not describe the business in any helpful way.

For the first half, revenue was $169,000 (versus $4,000), the net loss was $12.4 million (versus $10.2 million), and operating expenses were $13.1 million (versus $10.7 million, +22.9%).

Where the extra spending went

The 10-Q explains nearly every dollar of the cost increase, and very little of it was new research:

• General and administrative, +$1.1 million (+55%): per the MD&A, about $541,000 more bonus accrual, $298,000 more stock-based compensation, and $296,000 more "legal costs and filing fees related to our patent portfolio." For a licensing company the patent spending makes sense, since the patents are what it sells. The bonus and stock-pay increases are compensation, not new capability.
• Research and development, +$283,000 (+9%): $146,000 of higher bonus accrual, $96,000 of stock-based compensation and $40,000 of tool lease costs. Underlying lab activity grew only modestly.
• Selling and marketing, +$296,000 (+210%): $273,000 of higher employee costs and $70,000 of stock pay, "all related to an increase in headcount." This is the one cost line clearly aimed at turning engagements into signed deals.

Compared with Q1 2026, operating expenses rose from $6.2 million to $6.9 million, almost entirely in G&A ($2.3 million to $3.2 million). R&D actually fell slightly, from $3.46 million to $3.29 million.

What the headline numbers hide

• Flat EPS conceals a wider loss. The loss grew by $1.4 million, but weighted shares grew 27%, so the per-share loss stayed at $0.17. Shares outstanding rose from 32.4 million at December 31, 2025 to 39.0 million at June 30, 2026, a 21% increase in six months. Existing holders' slice of any future royalty stream got smaller by that much.
• Two very different share prices this year. In the first half, Atomera sold about 1.3 million shares through its at-the-market (ATM) program, meaning shares sold gradually into the open market, at an average of about $2.47 for $3.1 million net. In February it also sold 5 million shares to institutional investors at $5.00 for $23.6 million net. No ATM shares were sold in Q2. The $5.00 placement accounts for most of the new cash.
• Cash burn is lower than the loss, partly for timing reasons. First-half operating cash outflow was $8.5 million against a $12.4 million net loss. The gap comes mainly from $3.1 million of stock-based compensation, which costs no cash but dilutes shareholders, and a $468,000 rise in accrued payroll that the filing describes as "primarily... our annual bonus accrual." That bonus has been expensed but not yet paid, so it will become cash outflow later. Year-over-year cash burn was nearly flat ($8.5 million versus $8.3 million) even though the reported loss grew.
• Adjusted EBITDA excludes the fastest-growing cost. The company's non-GAAP adjusted EBITDA loss of $5.0 million removes stock-based compensation, which rose 36% to $1.74 million in the quarter. For a company that pays its people partly in shares and funds itself by issuing them, that cost is real.
• Investment income helped a little. Interest plus accretion income (the gain as discounted short-term securities approach maturity) was $350,000, up from $234,000, because the new cash was moved into short-term investments. Other income of $74,000 was mostly a state R&D tax credit, about the same as a year ago.
• No one-offs of note in Q2. Severance of about $52,000 appears only in the first-half R&D figures. Deferred revenue is zero, so there is no backlog of prepaid license fees waiting to be recognized.

Takeaway: Atomera's quarter was about runway, not revenue. The February share sale lifted cash to $38.4 million, enough for roughly two years at the current burn. But the $158,000 of revenue came from wafer deliveries, not licenses, and costs are rising faster than any sign of commercial traction. That makes an R&D or HVM license signing, not the quarterly loss, the figure that matters.

Outlook

Management gave no financial guidance. The earnings release points to three areas of customer activity: "continued strong progress with GAA customers" (gate-all-around, the newest transistor design used at leading-edge logic factories), "growing interest from DRAM and flash memory providers", and a new approach to gallium-nitride-on-silicon (GaN-on-Si) for radio-frequency chips. CEO Scott Bibaud described the GaN results as having "the potential to enable an entirely new class of RF devices." The 10-Q states that working capital is "sufficient to fund our presently forecasted working capital requirements for, at least, the next 12 months." It also says the company will need to raise more capital if license fees and royalties do not arrive in time, with the $50 million 2025 ATM program available for that.

Our view: The progress described is about engagements and interest, with no new license named. The money the filing shows coming in still comes from evaluation work. Atomera has spent years at this stage, as its $254 million accumulated deficit shows, so any new license signing should be the main thing judged in coming quarters. Cost growth is the second thing to watch. A 33% rise in operating expenses with no matching revenue shortens the runway the February raise just bought. If Q3 costs keep rising at that pace without a new R&D or HVM license, another round of share sales, more likely through the ATM, becomes probable within roughly 18 months. The Q3 10-Q is expected around early November 2026, based on the roughly 90-day filing cadence.

This is Atomera's first report on this site, so there is no earlier outlook to check against.