Ideal Power Inc. (IPWR) | The Buildout — AI Infrastructure

——Mkt cap — · 52-wk —–— · YTD — · delayed
Updated Oct 1, 2026Q2 FY2026 reviewed
Ideal Power designs B-TRAN solid-state switches aimed at circuit protection in 800-volt DC AI data center power systems.
Funnel >$400M
Stated sales opportunity funnel, up from about $300M at the May call.
Foundry signed
Long-term Asian supply deal; functional first silicon the same quarter.
Cash $41.3M
Q2 cash burn of $2.5M came in at the low end of guidance.
No revenue yet
Q2 2026 revenue rounds to $0.0M; FY2025 was $37,728.
The Buildout Takeaway
Ideal Power spent the quarter turning narrative into structure — a long-term foundry, a bigger stated funnel, a funded balance sheet — while reported revenue stayed near zero. The gap between those two facts is the story: B-TRAN is still in prototype evaluation with no field deployment, and by the company's own schedule the 800V DC architecture its products serve does not begin rolling out until the second half of 2027.
10 analysts·7 Buy3 Hold0 Sell
Coverage is thin — no price estimates on file, so no target is shown

FY2026 cash burn guidance approximately $10M-$10.5M · no revenue guidance given
Important: The Buildout is a data analytics platform. Content is generated by algorithms and AI agents using public filings, earnings transcripts, and market data. This is not personalized investment advice.
Our View

The Verdict

Ideal Power is a single-business, fabless power semiconductor company in Austin, Texas. It is developing one device: B-TRAN, a bidirectional transistor rated at 1200 volts that conducts and blocks current in both directions in a single switch, where silicon carbide MOSFETs and JFETs need at least two devices back-to-back. Its two launched products — a discrete B-TRAN die and the SymCool power module — are designed for solid-state circuit protection and electric vehicle contactors. The AI link runs through one architecture: the migration of data center power distribution from 400-volt AC to 800-volt DC, a shift NVIDIA has been driving through its GPU platform roadmap. That architecture requires solid-state circuit breakers for ultrafast fault handling, and B-TRAN is designed to be the switch inside them.

Market Cap—
Revenue (TTM)$0M
Net Cash$41M
Earnings Beats3 of 7
P/E (TTM)—
EV/EBITDA (TTM)—

What We Like

  • A long-term supply agreement with a high-volume, automotive-qualified wafer foundry in Asia — explicitly not China — that has built more than 1 billion power semiconductors; functional first silicon was achieved roughly a quarter after discussions began.
  • Existing foundry relationships carry capacity for 2-plus years, and the new agreement is tied to management's 40%-plus gross margin target at scale.
  • $41.3 million of cash at June 30, 2026, after a $27.7 million net registered direct offering led by the company's largest institutional shareholders; management said it gives partners confidence the company has adequate capital for several years.
  • 105 issued patents, with the double-sided wafer process flow held as a trade secret rather than patented — management says a competitor studying the patents would still lack the know-how to fabricate the device.
  • Two named commercial relationships: a late-2024 SSCB design win with one of the largest circuit protection equipment manufacturers in Asia, and a 2022 product development agreement with Stellantis for a custom B-TRAN power module.

What We’re Watching

  • Q4 2026 is the load-bearing window: lead Asia customer prototype orders, hyperscaler prototype delivery, the Stellantis project milestone, and industrial JEDEC qualification completion all land in the same quarter.
  • Both customer-controlled commitments moved in the same quarter: Stellantis deliverable timing was 'impacted,' shifting the next milestone from mid-2026 to Q4 2026, and automotive qualification went from 'this summer' to 'align with customer timelines.'
  • Funnel conversion: the more than $400 million sales opportunity funnel is a management-defined opportunity metric, not backlog, and management calls converting it 'our top priority.'
  • Industry timing: the 800V DC data center rollout those products serve is expected to start in the second half of 2027, with adoption into 2028 — well after the 2026 prototype milestones.
Bottom Line

The case strengthened structurally and stayed unproven commercially. In one quarter the company signed a long-term foundry agreement, lifted its stated funnel above $400 million, and took cash to $41.3 million, while reported revenue rounded to zero and both customer-controlled commitments moved. The pattern across the last two calls is a company that keeps the items it controls and reframes the ones it does not. The open question is whether the Q4 2026 prototypes convert into orders and eventually production revenue, on a timeline where the industry's 800V DC rollout does not begin until the second half of 2027.

Next upNext up is the Open Compute Project Global Summit in October 2026, where management and its industry partner introduce the intelligent SSCB prototype concept. Third-quarter results follow, then the Q4 2026 milestone cluster — Asia customer orders, hyperscaler prototype delivery, the Stellantis project milestone, and industrial qualification completion — which tests whether the funnel starts converting.
Last Quarter — Q2 FY2026

Earnings

Ideal Power reported effectively no revenue in the quarter ended June 30, 2026; the figure rounds to $0.0 million, consistent with management's description of 'modest revenue.' Gross margin was 30.9% on that minimal revenue, against -172.7% in the year-ago quarter. EBITDA was -$3.5 million and the net loss $3.4 million. Cash burn was $2.5 million, at the low end of the $2.5 million to $2.7 million guidance, and cash and equivalents stood at $41.3 million at June 30 after the May raise.

MetricQ2 FY2026Q1 FY2026Q2 FY2025YoY
Revenue$0M$0M$0M—
Gross margin30.9%—-172.7%+20360bps
EBITDA−$4M−$4M−$3M+16.7%
EPS$-0.20$-0.32$-0.33−39.0%
Cash burn$2.5M$2.3M$2.5Mflat
Sales opportunity funnel>$400M>$300Mn/a—
While growing funnel is encouraging, converting it into design wins, production orders and revenue remains our top priority.— David Somo, CEO, 2026-08-13

Management tone: The tone shifted on qualification posture and balance sheet. In May, management said it was 'making progress towards completing automotive reliability testing this summer'; in August it re-sequenced to industrial-first and said automotive testing 'will be planned to align with customer time lines.' On capital, May's call sat on $16.4 million of cash; August's sat on $41.3 million, with management describing capital as adequate for several years. Management was also direct that the strategy does not change because of the larger balance sheet, and it declined to put a date on the hyperscaler evaluation, saying the timing was 'difficult to call.'

Management Guidance

Full-year 2026 cash burn is guided to approximately $10 million to $10.5 million. Quarter-to-quarter variability in operating expenses is tied partly to the timing of semiconductor fabrication runs. No revenue guidance was given; the Q1 2026 filing said only that the company expects to recognize modest revenue from product sales and development agreements in the remainder of 2026.

Business Trajectory

Trajectory

Revenue has been effectively zero for several quarters: $0 in Q1 2026 and a figure rounding to $0.0 million in Q2 2026, against FY2025 commercial revenue of $37,728. Losses are steady rather than improving — EBITDA of -$3.5 million in Q2 2026 versus -$3.0 million a year earlier, and a net loss of $3.4 million versus $3.0 million. Gross margin was 30.9% in Q2 2026 against -172.7% in Q2 2025, on negligible revenue. What actually changed is the balance sheet and the forward pipeline: cash went from $16.4 million at March 31 to $41.3 million at June 30 after the $27.7 million net raise, and the stated sales opportunity funnel rose from about $300 million in May to over $400 million in August.

Revenue & Margin Trajectory
RevenueGross margin$0$0$0$1$0M$0M$0M$0M$0M$0M$0M$1M$0M−$1M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M-68%31%Q3'16Q4Q1'17Q2Q3Q4Q1'18Q2Q3Q4Q1'19Q2Q3Q4Q1'20Q2Q3Q4Q1'21Q2Q3Q4Q1'22Q2Q3Q4Q1'23Q2Q3Q4Q1'24Q2Q3Q4Q1'25Q2Q3Q4Q1'26Q2
RevenueGross margin$0$0$0$1$0M$0M$0M$0M$0M$0M$0M$1M$0M−$1M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M$0M-68%31%Q3'16Q4Q1'17Q2Q3Q4Q1'18Q2Q3Q4Q1'19Q2Q3Q4Q1'20Q2Q3Q4Q1'21Q2Q3Q4Q1'22Q2Q3Q4Q1'23Q2Q3Q4Q1'24Q2Q3Q4Q1'25Q2Q3Q4Q1'26Q2
Gross margin as reported.
Share Price — 12 Months
$2$5$8$052-wk high $8Oct '25DecMar '26JunOct '26
52-week range $3–$8.
Share Price — 12 Months
$2$5$8$052-wk high $8Oct '25DecMar '26JunOct '26
52-week range $3–$8.
The Numbers

The Model

The model projects FY+1 revenue of $0.2 million with EBITDA of -$14 million (-6700% margin), and FY+2 revenue of $2.0 million with EBITDA still -$14 million (-725% margin). Across the five independent runs the FY+2 revenue spread is 300%, ranging from $2 million to $8 million, so the projection is wide. The near term is anchored on sampling and development-kit orders and NRE fees, which is what management says near-term customer revenue consists of. FY+2 depends on whether the Q4 2026 prototypes convert into orders and whether the 800V DC data center rollout — which management expects to start in the second half of 2027, with adoption into 2028 — carries volume.

Revenue & EBITDA Projections
REVENUE$0M$0M$2MFY25FY+1 (E)FY+2 (E)EBITDA & MARGIN−$10M−$14M−$14MFY25FY+1 (E)FY+2 (E)
REVENUE$0M$0M$2MFY25FY+1 (E)FY+2 (E)EBITDA & MARGIN−$10M−$14M−$14MFY25FY+1 (E)FY+2 (E)
Solid bars are reported actuals; outlined bars are model projections — not company guidance.
MetricFY2025Next FY (E)Following FY (E)
Revenue$0M$0M$2M
YoY Growth——+900.0%
EBITDA−$10M−$14M−$14M
EBITDA Margin—-6700.0%-725.0%

Projections are the median of 5 independent model runs.

Full-year 2026 cash burn is guided to approximately $10 million to $10.5 million. Quarter-to-quarter variability in operating expenses is tied partly to the timing of semiconductor fabrication runs. No revenue guidance was given; the Q1 2026 filing said only that the company expects to recognize modest revenue from product sales and development agreements in the remainder of 2026.

What Could Go Right — and Wrong

What good looks like
  • The lead Asia customer places initial low-volume orders in Q4 2026, converting the stated funnel into contracted revenue for the first time.
  • The intelligent SSCB prototype is delivered for the hyperscaler's NVIDIA Rubin Ultra evaluation, the expected purchase order arrives, and follow-on evaluations with additional hyperscalers materialize.
  • Industrial JEDEC qualification completes in Q4 2026, unlocking the AI data center, energy storage, and grid opportunities management calls its nearest-term revenue.
  • Stellantis completes the remaining purchase-order deliverables on the Q4 2026 milestone and moves toward a production decision on the EV contactor program.
  • Volume orders ship from the new foundry and gross margin approaches the 40%-plus target tied to that foundry's cost structure.
What could go wrong
  • The funnel does not convert: no production order arrives, and reported revenue stays near zero.
  • Stellantis deliverables slip again past Q4 2026, or automotive qualification remains undated.
  • The 800V DC data center rollout slips materially past the second half of 2027, pushing volume revenue into 2028 and beyond.
  • A competitor or a system integrator's in-house solution takes the socket — silicon carbide or JFET switches, or integrator programs such as Eaton's acquired Resilient Power, ABB's Advantics, or Vertiv's own solid-state development.
  • Cash burn overshoots the FY2026 guide of $10.3 million to $10.5 million, or the company draws on its shelf registration.
What’s Next

Looking Ahead

The next twelve months run through a dense Q4 2026 cluster: initial low-volume orders for the lead Asia customer's prototype builds, delivery of the intelligent SSCB prototype for the hyperscaler's NVIDIA Rubin Ultra evaluation, the next Stellantis project milestone, and completion of industrial JEDEC qualification — with the low-current SSCB prototype having shipped to the Asia customer in August and the intelligent SSCB concept introduced at the Open Compute Project Global Summit in October 2026. Every forward commitment in the record is a development, prototype, qualification, or supply-chain milestone rather than a contracted order, and no backlog value has been disclosed. The industry rollout those milestones feed is expected to start in the second half of 2027.

Catalysts
  • August 2026Asia prototype shipment — Low-current SSCB prototype shipped to the lead Asia customer for testing.
  • October 2026OCP Summit introduction — Intelligent SSCB prototype concept introduced with the industry partner.
  • Q4 2026Asia initial orders — Initial low-volume orders to support the customer's prototype builds.
  • Q4 2026Hyperscaler prototype delivery — Intelligent SSCB prototype due for NVIDIA Rubin Ultra evaluation.
  • Q4 2026Industrial qualification — JEDEC qualification completes, gating near-term AI and grid revenue.
  • Q4 2026Stellantis milestone — Next project milestone supported by completed PO deliverables.
Numbers

Financials

Annual Summary

MetricFY2024FY2025TTMYoY
Revenue$0M$0M$0M-100.0%
Gross Margin-345.5%-112.1%—+23,339bps
EBITDA−$11M−$10M−$12M+2.8%
EBITDA Margin-10700.0%———
Net Income−$10M−$10M−$12M+0.0%
Free Cash Flow−$9M−$9M−$9M—
Net Cash————

Key Ratios (Trailing)

Valuation
  • P/E TTM—
  • EV/EBITDA TTM—
  • EV/Revenue TTM—
  • Price/FCF TTM—
Profitability
  • ROIC-602.4%
Reference

The Company

Ideal Power makes one thing: the B-TRAN bidirectional switch. Unlike silicon carbide MOSFETs and JFETs, which need at least two devices back-to-back to conduct and block in both directions, a single B-TRAN does both, is rated at 1200 volts, and handles faults in microseconds. The company has launched two commercial products built on it — a discrete packaged B-TRAN die and the multi-die SymCool power module — both designed to meet the very low conduction loss needs of solid-state circuit protection and electric vehicle contactor markets. It has extended to the system level with solid-state circuit breakers, an intelligent SSCB prototype, SSCB reference design kits aimed at incumbent electromechanical breaker makers, and earliest-stage solid-state transformers. The AI relevance is the 800-volt DC power distribution that next-generation AI racks require, where solid-state breakers provide the ultrafast fault handling.

Ideal Power is fabless. Wafer fabrication and packaging are outsourced, and the counterparties are deliberately unnamed: management treats the double-sided wafer process flow as a trade secret and says that even a competitor studying its patents would lack the know-how to fabricate the device. Foundries and packaging houses are chosen exclusively in countries the company says have a history of respecting intellectual property rights, and customers can learn the foundry's identity under NDA. Its only disclosed facility is a leased 5,775 square-foot office and laboratory in Austin, Texas, and it reports a single operating segment. Operations have been funded primarily through the sale of common stock and pre-funded warrants. It holds 105 issued patents and has added a Europe sales director, applications engineering support in Asia, and a quality hire supporting automotive qualification.

Business Segments

discrete B-TRAN®
Single packaged B-TRAN die
Designed for the very low conduction loss needs of solid-state circuit protection and EV contactor markets.
Growth driver: SSCB and EV contactor demand
SymCool® Power Module
Multi-die B-TRAN module
Same target markets as the discrete device: solid-state circuit protection and EV contactors.
Growth driver: Higher-current circuit protection designs
System-level products
Low- and medium-current SSCBs; earliest-stage solid-state transformers
Solid-state circuit breakers, an intelligent SSCB prototype, and reference design kits aimed at incumbent breaker makers.
Growth driver: 800V DC data center circuit protection

Competitive Landscape

The 10-K lists the leading players in the global power semiconductor market as competitors: Infineon Technologies, Texas Instruments, onsemi, Fuji Electric, STMicroelectronics, Mitsubishi Electric, Rohm Semiconductor, and Renesas Electronics. Management's differentiation is technical rather than commercial: B-TRAN conducts and blocks in both directions in a single 1200-volt device, while JFETs and silicon carbide MOSFETs need at least two devices in back-to-back configuration. No design win against silicon carbide appears in the record. The second source of competitive pressure comes from the system integrators buying or building the capability themselves — Eaton acquired Resilient Power for medium-voltage solid-state transformers, ABB acquired Advantics for DC and silicon-carbide technology, and Vertiv describes its own solid-state work as project development.

  • Infineon Technologies AG
    Named in the 10-K's leading-players list; also referenced in Q&A as the maker of a CoolSiC JFET comparison.
  • Texas Instruments
    Named in filings; not discussed.
  • onsemi
    Named in filings; not discussed.
  • Fuji Electric
    Named in filings; not discussed.
  • STMicroelectronics
    Named in filings; not discussed.
Competitor rows come from the FY2025 10-K's list of leading power semiconductor players; the Infineon CoolSiC JFET reference is from the Q2 2026 call Q&A.

Supply Chain

Ideal Power sits upstream of circuit-protection OEMs and contactor makers: it designs the switch, outsources wafers and packaging, and sells into breakers and contactors that eventually reach data centers. No neighbor transcript in the source material mentions Ideal Power by name.

Supplier
Unnamed high-volume Asian foundry (not China)
Wafer fabrication; automotive-qualified, more than 1 billion power semiconductors built
Supplier
Existing wafer foundries (unnamed)
Wafer fabrication; capacity for 2-plus years per management
Supplier
Packaging partners (unnamed)
Device packaging
→
Trade-secret process know-how
IPWR
Fabless; a single 5,775 sq ft Austin office and laboratory, no owned fab.
→
Stellantis
Custom B-TRAN power module for EV drivetrain inverters
Lead Asia circuit-protection manufacturer (unnamed)
SSCB design win from late 2024; low- and medium-current SSCB programs
U.S. hyperscaler (unnamed, via industry partner)
Intelligent SSCB prototype for NVIDIA Rubin Ultra 800V DC evaluation

Analysis updated Oct 1, 2026, reviewing Q2 FY2026. Prices delayed. Built with The Buildout’s published methodology. Not investment advice. No positions held. © The Buildout 2026.

More on IPWR: Earnings recap