GRDX — Q2 2026 Financial Report Analysis

A $10.1 million goodwill write-off nine months after the deal that created it

GridAI Technologies Corp. — the company formerly known as Entero Therapeutics, and before that First Wave BioPharma — reported a net loss of $13,040,293 for the three months ended June 30, 2026, against a loss of $998,271 in the same quarter a year earlier. Roughly 77% of that came from a single non-cash line: a $10,070,124 goodwill impairment charge written against the acquisition of Grid AI Corp., which the company closed on September 30, 2025.

Goodwill is the premium an acquirer pays above the identifiable value of what it buys — the part of the purchase price that only holds up if the acquired business performs as projected. GridAI recorded $22,434,494 of it on the Grid AI deal, against a total consideration of $32,397,116 (Note 3). Within three quarters, it wrote off 45% of that premium. The filing's stated reason is specific: "actual operating performance and updated near-term financial projections for the Grid AI business were significantly below the projections established at the acquisition date, primarily due to delays in commercialization and customer contract execution."

Revenue for the quarter was $85,876. Cost of services to produce it was $122,390.

The quarter in numbers

Prior-year comparisons here are close to meaningless in the usual sense: in Q2 2025 this was a clinical-stage gastrointestinal drug developer with no revenue and no energy business. Grid AI has only been consolidated since September 30, 2025, so "year-over-year growth" is really the arrival of a different company inside the same listing.

The more useful comparison is sequential. Six-month revenue was $124,084 against six-month cost of services of $774,338, which means Q1 2026 produced $38,208 of revenue on $651,948 of direct cost. Management's own explanation, in plain terms, is that the work and the billing do not line up in time: "The quarterly pattern reflects the timing and mix of project and service activity, and direct service costs did not vary proportionately with quarterly revenue." That is a normal feature of engineering and construction contracts — which is what a meaningful part of this revenue is. The revenue recognition policy (Note 2) covers energy generation under power purchase agreements, engineering, procurement and construction ("EPC") services recognized over time on a cost-to-cost basis, and operations-and-maintenance fees. It is project revenue, not software subscription revenue, and it is currently running below its own direct costs.

What the impairment test actually disclosed

This is where the amended filing earns its place. Under the accounting rules, when something signals that an acquired business may be worth less than its carrying amount, the acquirer must estimate the unit's fair value and write goodwill down by the shortfall. GridAI's numbers, disclosed in the amendment:

A 42.5% discount rate is the number to sit with. The discount rate is the annual return a buyer would demand for taking on this business's risk; it is applied to projected future cash flows to work out what they are worth today. Rates in the low 40s are what valuation practitioners use for venture-stage assets whose cash flows may never arrive. Management's own valuation therefore treats the AI unit as early-stage and high-risk — consistent with the filing's explanation that market-based comparisons were rejected because "the AI reporting unit is in an early stage of development and was projected to incur near-term operating losses."

Note also what did not move: the discount rate came down from 45.0% to 42.5%, which by itself raises fair value. The entire impairment therefore comes from cutting the forecast, not from a harsher view of risk. Management is explicit that it revised "near-term and intermediate-period revenue and cash flow expectations downward and delayed the expected profitability ramp" while "longer-term expectations remained substantially consistent," and frames the write-down as "a delay in the timing of projected revenue and cash flows, rather than a fundamental change in the expected long-term performance."

That framing deserves scepticism, for an arithmetic reason. Pushing cash flows out in time, at a 42.5% discount rate, destroys value extremely fast — a one-year delay alone cuts the present value of a given cash flow by roughly 30%. A $10.07 million write-down is consistent with a delay of that order; it does not require the long-term case to have been abandoned. But the corollary is that at this discount rate, one more year of slipped contracts would take out a similar amount again, and $12.4 million of goodwill remains on the AI unit with $20.7 million of related finite-life intangible assets sitting alongside it. The filing discloses no impairment charge against those intangibles this quarter, so they remain fully exposed if the forecast slips further.

About the source filing: a 10-Q/A driven by SEC staff comments

The quarter was originally reported in a Form 10-Q filed August 21, 2026 (accession 0001104659-26-099751). On September 18, 2026 the company filed Amendment No. 1 on Form 10-Q/A (accession 0001104659-26-108635). The explanatory note states the amendment revises goodwill disclosures in Note 3 and Note 7 and expands the critical accounting estimates discussion "in response to comments received from the staff of the SEC," and that the revisions "do not change any amounts reported in the Company's condensed consolidated financial statements, the goodwill impairment charge recognized as of June 30, 2026, or any other previously reported financial results."

So this was not a restatement — no reported figure changed, and every number in this analysis is identical in both versions. But the amendment is not cosmetic either, and one change is worth reading closely. The original filing said the Grid AI goodwill was "primarily attributable to expected synergies from integrating Grid AI's operations with the Company's existing platform, including anticipated growth in AI-driven energy optimization solutions, expansion into new markets, workforce expertise, and assembled workforce." The amended text withdraws that rationale:

Grid AI Corp. continues to operate as a separate operating business, and the Company does not currently anticipate significant operational integration between its AI business and its legacy biopharmaceutical business. Any anticipated efficiencies from the transaction are expected to result primarily from shared corporate infrastructure rather than the integration of operating activities.

In substance: the $22.4 million premium was not, after all, supported by combining an energy-software business with a gastrointestinal drug programme. What is left as justification is shared corporate overhead, management resources and the assembled workforce. That is a considerably thinner economic story for a premium of that size, and it is the sort of walk-back that a securities regulator's review tends to produce rather than a voluntary filing.

Two businesses that share almost nothing but a balance sheet

GridAI reports two segments. The AI segment is Grid AI Corp. and its subsidiaries, including AMPX UK Holdings and AMPX Limited, working on distributed energy resource optimisation and grid-edge applications. The GI segment is the leftover biopharmaceutical business, now reduced to a single asset: Adrulipase, for exocrine pancreatic insufficiency. The Latiglutenase and CypCel programmes left with the ImmunogenX rescission completed December 31, 2025; the Capeserod licence from Sanofi was terminated; Niclosamide is no longer pursued.

In the second quarter alone, the AI segment lost $10,872,231 at the operating line (almost entirely the impairment) and the GI segment lost $2,234,529. Total segment assets at June 30 were $33,776,129 for AI and $3,259,471 for GI.

Research and development spending tells its own story about the legacy business: $8,424 for the quarter. For a company whose remaining drug asset failed its primary efficacy endpoint in a July 2023 Phase 2b study, that is not a development programme in any active sense — the filing says next steps remain "subject to available capital and strategic priorities." Meanwhile general and administrative expense ran at $2,991,698 in the quarter, about 35 times the entire R&D line and roughly 35 times revenue.

The balance sheet is almost entirely the acquisition

Goodwill plus acquired intangibles come to $34,736,416 — 93.8% of total assets. Strip them out and stockholders' equity of $14.9 million becomes negative tangible equity of roughly $19.9 million. Set against that, current assets of $2.3 million stand against current liabilities of $14.1 million: a working capital deficit of about $11.8 million, which the company states directly in its going-concern note.

Inside those current liabilities sits $6,250,000 of deferred consideration — the unpaid balance of the $7,000,000 the company still owes for Grid AI itself (Note 15; $750,000 was paid during 2025). It is non-interest-bearing and, per the filing, "expected to be settled through future cash payments or other negotiated arrangements." The company has therefore written off $10.1 million of the value of an acquisition it has not finished paying for.

One line deserves flagging because it is easy to misread as cash raised: shareholder and subscription receivables of $4,363,499, shown as a deduction from equity (it was zero at December 31, 2025). The May 2026 private placements were structured so that, on the larger $5.25 million tranche, "20% of the gross proceeds was payable at closing and the remaining 80% is payable upon effectiveness of the related Form S-1 resale registration statement" — roughly $4.2 million contingent on a registration statement going effective. The equity is booked; a large part of the money is not yet in the bank.

Cash: burning about $550,000 a month, funded entirely by issuing stock

Operating cash burn of $3.29 million over six months is about $550,000 a month, against $1.38 million of cash at quarter end — under three months of runway on the quarter's own figures. Note that the $16.5 million six-month net loss converts to only $3.3 million of cash outflow because $10.1 million of it was the impairment, plus roughly $1.1 million of intangible amortisation, $1.3 million of stock-based compensation and $0.7 million of debt discount amortisation. The loss is mostly non-cash; the burn is real but far smaller.

Financing inflows of $4.14 million comprised roughly $3.3 million from the May private placements, $0.6 million from warrant exercises and $1.0 million from promissory notes, less $0.7 million repaying the revolving loan principal. That revolver — an 18% facility from 1396974 BC Ltd. that the company failed to repay at its January 31, 2026 maturity, triggering an April 1 demand for $1,014,675 — was settled in May for $800,000 cash plus 71,482 shares at a deemed $3.25. The facility is now closed to further borrowing.

The going-concern language is unusually blunt for a company with a live acquisition story. Management states that its plans "do not alleviate substantial doubt," and lists among the options under evaluation "a merger, reverse merger, sale, wind-down, liquidation and dissolution or other strategic transaction," with failure potentially forcing it to "seek protection under the provisions of the U.S. Bankruptcy Code."

What happened after the quarter closed — including $2.96 million lent out

Three post-quarter items matter more than the quarter itself.

A larger, higher-priced raise. On July 1, 2026 the company agreed to sell 108,932 shares, pre-funded warrants over 1,742,991 shares and common warrants over 1,851,923 shares for gross proceeds of about $8.5 million, at $4.59 per unit, closing July 7. That is a marked improvement on the May pricing of $2.00 and $2.865 per unit. It is also up to 3.7 million additional shares against 6.9 million outstanding at June 30 — potential dilution of more than half the share count, on top of the 63.7% increase already booked in the first half.

Money going out the door to a third party. On July 17, 2026 GridAI lent $2,000,000 to Pronghorn Resources, LLC under a secured convertible note at 7%, due the earlier of December 31, 2026 or a change of control. A subsequent Form 8-K (filed September 11, 2026, event dated September 4) discloses an amended and restated note reflecting a second loan of $964,000, taking the total to $2,964,000. On default, GridAI may convert into membership interests equal to 10% of Pronghorn's fully diluted equity. The filing notes that Nicholas D'Onofrio, a Pronghorn shareholder, "also holds common stock and common warrants of the Company."

Put those two together and the capital allocation is striking: a company that told investors in the same document that substantial doubt about its survival was unalleviated raised roughly $8.5 million gross and, within ten days, began lending out about a third of it to an unrelated resources company in which one of its own shareholders has an interest. Neither the 10-Q nor the 8-K explains how the loan advances the energy-technology strategy.

Legal and control matters. WestPark Capital filed a JAMS arbitration on May 26, 2026 seeking approximately $2,057,600 over an alleged breach of a right of first refusal; the company denies it and has accrued $202,000. A $209,386 Asymchem judgment, which had caused a bank account restraint, was settled by stipulation on August 17, 2026. Separately, management concluded that disclosure controls and procedures "were not effective" at June 30, 2026 "due to the existence of a material weakness in internal control over financial reporting related to the integration of Grid AI Corp., including insufficient accounting resources and processes." That admission sits alongside a pattern: the company filed a late-filing notice (Form NT) before both its Q1 and Q2 2026 quarterlies and before its FY2025 annual report, and the annual report's delay briefly put it out of Nasdaq compliance in April 2026.

Takeaway: The reported loss is dominated by a non-cash write-down, but the write-down is the real news — nine months after paying $32.4 million for Grid AI, GridAI has erased 45% of the resulting goodwill, told the SEC that the acquisition synergies it originally cited do not exist, and valued the acquired business using a 42.5% discount rate that treats it as venture-stage. With $124,084 of six-month revenue earned at negative gross margin, $12.4 million of goodwill and $20.7 million of intangibles still carried against it, and $6.25 million of the purchase price still unpaid, the central question is not whether the AI business grows but whether the remaining acquisition carrying value survives the next forecast revision.

What to watch next

Management gave no revenue or earnings guidance — unsurprising for a company at this stage — but it did commit to a view: that the commercial opportunity is intact and only the timing has moved, and that Grid AI "remains central to the Company's go-forward strategy." That claim is testable, and the tests arrive quickly.

The first is the Q3 2026 10-Q, due in November. Three things in it will matter more than the headline loss: whether AI-segment revenue moves beyond the five-figure range and whether it is finally earned above its direct costs; whether any further goodwill or intangible impairment appears; and whether the purchase price allocation, which the filing says stays preliminary through the September 30, 2026 measurement-period close, is adjusted. That last one falls in the third quarter, so the Q3 filing should settle it.

The second is funding. Even after the July raise, the structural position is unchanged: an operating burn around $550,000 a month, a working capital deficit near $11.8 million, $6.25 million of deferred acquisition consideration sitting in current liabilities, and roughly $3.0 million lent out to Pronghorn and not due back until December 31, 2026 at the earliest. Recovery of that loan on time is now a genuine liquidity item, not a footnote. Continued equity issuance at each cash squeeze is the base case, and the warrant overhang from the May and July rounds means the share count can expand substantially without a single new deal being announced.

The third, and least visible, is disclosure quality. The SEC's staff has already required this company to rewrite why it recognised $22.4 million of goodwill, and management has conceded a material weakness tied to accounting for the same acquisition. Until that weakness is remediated, every estimate in these statements — the reporting unit's fair value above all — carries more uncertainty than the precision of the figures suggests.

Source: GridAI Technologies Corp. Form 10-Q/A (Amendment No. 1) for the quarterly period ended June 30, 2026, filed with the SEC on September 18, 2026 (accession 0001104659-26-108635), which amends the original Form 10-Q filed August 21, 2026 (accession 0001104659-26-099751); and Form 8-K filed September 11, 2026 (accession 0001104659-26-107100). The amendment revised goodwill disclosure only and changed no reported amounts.