A-Portfolio

The company in one breath

TurboGen, ticker TRBG on the Nasdaq Capital Market, builds small turbines that make electricity and useful heat where a building consumes them. Apartment owners, hotels, nursing homes, offices, and small data centers are the intended customers, but nobody has produced reported revenue for TurboGen yet.

The company wants customers to buy a machine or pay for energy as a service, which means TurboGen would own the equipment and collect money as customers use the power. Good would look like recurring energy payments, low maintenance, and enough customer savings to make switching obvious.

Why now

TurboGen began its Nasdaq dual listing on August 31 while keeping its Tel Aviv listing. A dual listing puts the same company on two exchanges. Good would look like deeper trading, better access to capital, and more commercial credibility.

This was not a normal American IPO. TurboGen registered 2.07 million existing shares for resale, and the company receives none of that selling money. Good would look like the public listing helping TurboGen win customers before existing holders use it mainly as an exit door.

The stock closed September 9 at $4.45, then showed $4.4698 before the September 10 open at 7:12 a.m. Eastern, according to Nasdaq. Yahoo Finance shows the first completed daily bar on September 3 closed at $11.84, so the stock lost 62.4% in four sessions. Good would look like several quiet weeks above a stable floor. Right now, the chart looks like somebody dropped a turbine off the roof.

TurboGen says three TR3200 systems should reach installation in the fourth quarter. That gives us a near-term test. Good would look like all three running at customer sites, passing certification, and producing the company's first reported sales.

The numbers, translated

• Revenue was $0 through December 31, 2025. Revenue is customer money that the accounting rules let a company count after it delivers. Zero means every valuation rests on machines and contracts that still need to become a business. Good would look like at least $2 million of recognized revenue over the next twelve months, then steady quarterly growth.
• The 2025 net loss was $19.5 million, up from $3.0 million. Net loss is everything the company spent or lost after all accounting charges. Most of the jump came from stock compensation and warrant accounting, but owners still absorbed the bill. Good would look like the annual loss falling below $10 million while revenue starts.
• Operating cash burn was $4.6 million in 2025, up 165%. Cash burn is real money that left during normal operations, so it matters more than the noisier paper loss. Good would look like less than $5 million of annual burn after production starts, with each new sales dollar consuming less cash.
• Cash was $3.9 million at year-end, and the filing later described a $7.3 million cash balance. That is the fuel available to finish products and start manufacturing. Management says the private placement and warrant proceeds can fund roughly 24 months, but manufacturing can eat forecasts for breakfast. Good would look like at least 18 months of cash after the first installations and no emergency share sale.
• Preliminary U.S. sales agreements covered 12 systems worth up to $5.4 million. Preliminary means the parties still needed final agreements, and the planned $90,000 of advances could be refunded under stated conditions. Good would look like signed final contracts, non-refundable deposits, delivery dates, and cash collected.
• TurboGen reported 23.39 million shares outstanding on August 21. At the $4.45 September 9 close, that puts the stock-market value near $104.1 million. Good for a company with no revenue would be impossible to define from sales, because dividing by zero is not valuation work, it is performance art. I need commercial proof before any multiple deserves respect.
• Options, restricted stock, and warrants could add about 2.07 million shares, equal to 8.9% of the reported count. Dilution means every current share owns a smaller piece when new shares appear. Good would look like revenue and cash value growing much faster than the share count.

What I like

The problem is real. Power demand from data centers, electric vehicles, and aging grids keeps climbing. Generating electricity beside the customer can reduce transmission strain and turn waste heat into hot water instead of warming the parking lot.

The product plan has range. TurboGen is working on systems from roughly 20 to 100 kilowatts, a kilowatt being a measure of how much power a machine can produce at once. Good would look like one design working reliably before management fills the menu with five more.

TurboGen claims its larger system can reach 33% electrical efficiency, versus 31% for comparable rivals. Efficiency tells you how much fuel becomes useful electricity instead of waste. Good would look like an independent test confirming the claim under real building loads, plus total efficiency above 80% after useful heat gets counted.

The energy-as-a-service model could create repeat payments. TurboGen claims customer savings of up to 15% versus grid rates. Good would look like signed customers saving at least 10% after fuel, maintenance, financing, and downtime, not just on a spreadsheet that never gets cold.

What worries me

TurboGen is still a development-stage company. It has demonstration production, no reported revenue, a going-concern warning, and customer agreements that can still dissolve. That is not a turnaround. That is a science project asking the public market to pay tuition.

Selling holders can move 2.07 million registered shares, about 8.8% of the August share count, and TurboGen gets no cash from those sales. Good would look like orderly volume and insiders holding because commercial progress improves the value, not because buyers cannot find the sell button.

Stock compensation reached $6.1 million in 2025, about 88% of general and administrative expense. Stock compensation pays people with ownership instead of cash. Good would look like this expense falling below 20% of operating costs and tying future awards to revenue, margins, and customer uptime.

Management announced former New York City mayor Eric Adams as an adviser on September 9 to help with introductions and media. Connections can open doors. They do not turn turbines, sign procurement contracts, or collect invoices. Good would look like named customers and contract values, not another photograph beside a podium.

The 30-day community scan found no useful on-topic Reddit, YouTube, Hacker News, GitHub, or prediction-market discussion. That means no independent crowd has tested the story in public. Good would look like customers and engineers debating operating results rather than broad energy chatter that never mentions the company.

What would make this a buy

I need price and proof together. The price must fall below $2.00, which would value the August share count near $46.8 million. That is not automatically cheap. It merely stops charging more than $100 million for a company with zero reported sales.

The proof requires all three planned fourth-quarter installations, at least $2 million of trailing revenue, gross margin above 25%, and twelve months of customer uptime data. Gross margin is the money left from each sales dollar after building and servicing the equipment. Good would look like TurboGen keeping at least 25 cents per dollar while customers still save money.

I also want cash above $8 million, annualized operating burn below $5 million, and no new financing below $2.00. This is research and commentary, never a buy signal. Until the machine makes revenue, waiting costs less than believing.

What makes me walk away

I walk if the three fourth-quarter installations slip past March 2027 without a precise explanation. A delay would say the move from demonstration to commercial hardware remains harder than management admits.

I walk if the preliminary U.S. agreements fail to become final contracts, or if customers demand refunds. Interest is not demand. A refundable deposit is not a moat.

I walk if cash drops below $4 million before quarterly revenue reaches $1 million, or if TurboGen sells more than 10% new shares before commercial proof arrives. That would make shareholders the product.

I also walk if independent testing cannot confirm the efficiency and maintenance claims. A one-point advantage over established competition is thin. If the test bench says no, the pitch goes in the trash.

IPO Watch

Holtec Nuclear (private, proposed ticker HNUC), expected September 18, range $15 to $18. Holtec designs nuclear equipment, stores spent fuel, decommissions old plants, and wants to build small modular reactors. The range tells us the proposed price per share, but the final valuation needs the share count too. I would write it up if the mature service businesses generate enough cash to fund reactor development without making new owners the permanent ATM.

Earnings On Deck

Adobe (ADBE), September 10 after the close. Watch whether artificial-intelligence tools lift recurring revenue without crushing margins, the portion of sales left after costs. A good print means customers pay more for the new tools and Adobe protects profit. A bad print means usage grows while pricing power fades, which matters to past feature Figma (FIG).

Reformation (REF), September 10 after the close. Watch sales growth, full-price demand, and inventory. Good means customers buy without heavy discounts and the company keeps healthy gross margin. Bad means markdowns clear clothes while clearing out profit too.

The verdict

Score: 30 out of 100. Bucket: pass. TurboGen earned points for a real grid problem, a sensible combined-power-and-heat design, planned installations, preliminary customer interest, and a repeat-payment model that could work. It lost far more for zero revenue, a going-concern warning, thin cash, refundable preliminary deals, heavy stock compensation, dilution, no independent proof, and a stock that fell 62.4% in four sessions.

A candidate starts at 80. TurboGen is fifty points short. I want to see the machines run. I do not want to pay owners more than $100 million while the income statement still reads zero.

The DBTech read

A preliminary sales agreement says two sides intend to do business, but important conditions can still kill the deal. TurboGen's two U.S. agreements cover up to $5.4 million, yet they still need final contracts and allow refunds under stated conditions. Revenue arrives only after TurboGen performs enough work to earn it under accounting rules. Good investors separate interest, orders, deliveries, and revenue because each step carries more proof. A menu order is not dinner until the plate hits the table.

Sources and as-of date

• Price: $4.45 at the regular close on September 9, 2026, and $4.4698 premarket at 7:12 a.m. Eastern on September 10, 2026, Nasdaq market data. The regular close drives the market-value calculation.
• Trading history: $11.84 September 3 close, $33.99 September 3 intraday high, and daily volume, Yahoo Finance chart data retrieved September 10, 2026 at 7:12 a.m. Eastern.
• TurboGen Form F-1/A, filed August 27, 2026. Business model, zero revenue, losses, cash, burn, pre-sales, supplier order, private placement, share count, dilution, competition, and going-concern risk.
• TurboGen listing Form 6-K, filed August 31, 2026. Nasdaq trading date and dual-listing status.
• TurboGen advisory-board Form 6-K, filed September 9, 2026. Eric Adams engagement and stated commercialization role.
• TurboGen official site, checked September 10, 2026. Current on-site power, combined heat and power, fuel flexibility, service intervals, and energy-as-a-service positioning.
• Nasdaq September IPO calendar, updated September 9, 2026. TurboGen resale listing and Holtec Nuclear expected date and range.
• Nasdaq earnings calendar, checked September 10, 2026. Adobe and Reformation report timing.
• Community research covered August 11 through September 10, 2026. It returned 23 broad Reddit threads but no item that actually discussed TurboGen, plus zero useful YouTube, Hacker News, GitHub, or Polymarket results. I treated the social sample as no signal, not evidence.
• Calculations use 23,389,747 shares, the $4.45 close, and Decimal arithmetic. Research and commentary only, never a buy signal.