Open in interactive viewer → charts, metric popovers & call review

Ideal Power: A $450M Funnel, a $36M Market Cap, and the Order That Slipped

The 800V DC story gets louder every quarter while the stock halves — and one Asian customer just moved the goalposts from 63 amps to 125.
IPWR · Earnings Call · 2026-11-12

The gap between the story and the tape

Ideal Power is a pre-revenue semiconductor micro-cap — roughly $36 million of market value — selling a bidirectional switch called B-TRAN into what may be the single most crowded narrative in the market today: the electrification of AI data centers. On paper, the third quarter was its best ever. The sales funnel grew to about $450 million from roughly $400 million three months earlier. A new partnership with Power Integrations landed. Reference design kits began shipping. And yet the stock is down 52.7% over the last 90 trading days, sitting 58% below its August 14 peak of $5.23. That divergence — a company narrating acceleration into a tape that says otherwise — is the whole story here.

What actually changed: the honest paragraph

Every earnings call has a section management would rather skip. This one's is unusually blunt. One year into the job, CEO David Somo opened with what he called "what has to be better," and buried in it is the quarter's real news:

Customer programs have taken longer to convert to production orders than expected. I remain confident that they will happen, but we need to accelerate the conversion of these opportunities into design-ins and revenue.

David Somo, President and Chief Executive Officer · 2026-11-12
The concrete casualty is the lead Asian customer that had been the nearest-term proof point — the one that took delivery of 63-amp SSCB prototypes and, per prior calls, was heading toward a small but symbolic fourth-quarter purchase order. It pivoted: “we no longer expect initial low-volume 63-amp prototype orders from this customer in the fourth quarter” — David Somo, President and Chief Executive Officer · 2026-11-12 To be fair to management, the pivot is not a performance failure — the prototypes met spec, and the customer simply raised its target to a 125-amp design after a new industry paper reshaped end-customer requirements. That distinction matters. But it also illustrates precisely the gating problem analysts have been circling for a year: these customers are electromechanical breaker makers navigating a moving architectural target, and reference design kits are the company's answer to speeding them up. That is why the RDK rollout — stocked at distributor Richardson Electronics, with more distributors expected — is quietly the most operationally important thing in the release, even if it produces almost no revenue.

Riding the 800-volt wave — with a validation partner

The 800-volt DC migration is not Ideal Power's invention; it is an industry-wide shift tied to the next generation of NVIDIA racks rolling out from the second half of 2027. Somo was explicit that he saw no deceleration: “we do not see any slowdown in investment in data centers” — David Somo, President and Chief Executive Officer · 2026-11-12 — and he pointed to peer power-semiconductor suppliers going on allocation as corroboration. This is a company riding a genuine macro wave, not manufacturing one, and the circuit protection niche it targets is where higher-voltage DC systems genuinely need faster, lower-loss interruption. The Power Integrations collaboration is the strongest third-party signal in the quarter, and analyst Casey Ryan read it the same way, calling it "a really strong market validation." The logic is sound: a self-powered intelligent breaker that derives its own low-voltage control power off the 800V bus eliminates an auxiliary supply, and it hands Ideal Power a partner with far more sales and applications engineering reach than it will ever build alone. The subtle change is in sequencing. The hyperscaler engagement — long framed as a hardware prototype — is now an engineering study first, NRE-funded, starting this quarter. Management spun this as a positive ("positions us to participate directly in the architectural decisions"), and it may be. But note the keyword drift: AI data center and energy infrastructure revenue is now firmly a second-half-2027 event, and CFO Tim Burns was careful to reset expectations about what comes first: “it's really the second half of next year where we're expecting to see the product revenue from the data center market” — Timothy Burns, Chief Financial Officer · 2026-11-12. The value between now and then is NRE agreements and RDK orders — not product.

The clock and the cash

What keeps this from being a going-concern story is the balance sheet. At September 30 the company held $38.3 million in cash, zero debt, and burned $2.9 million in the quarter — inside guidance. That leaves a cash runway measured in years rather than quarters, a direct consequence of the recent capital raise and the reason liabilities sit at barely 5.7% of assets. The income statement is the usual pre-revenue shape. Operating income ran at roughly -$4 million and net income at about -$3 million, both modestly worse year-over-year as headcount grew. Management guided full-year 2026 burn to $10.4–10.6 million versus $9.6 million in 2025, and flagged rising stock-based compensation as a source of quarter-to-quarter noise. One encouraging detail: R&D fell 25% year-over-year as the company rationalized its patent portfolio (108 issued patents unaffected).

Why this one matters

The tension is the trade. A company one year into a new CEO, with a $450 million funnel and a marquee validation partner, watched its stock halve on the same news cycle it announced the partnership. Prior calls brimmed with confidence — in May the message was “we are still on track with what we said in the business update call... there is no change in our expectation” — David Somo, President and Chief Executive Officer · 2026-05-14 on revenue timing, and in August management listed three separate reasons it was confident about commercialization. The current call is the first to concede that conversion is the bottleneck, not the opportunity. That is a more honest company — and a harder one to own, because the next visible catalyst is an NRE agreement, and the JEDEC qualification that unlocks volume is still a year-end promise rather than a delivered fact. The funnel can keep growing; until an order converts at the new 125-amp spec, the tape has decided to discount the arithmetic.