Intchains: The First Half’s “Revenue” Was a Related-Party Sale, and 2027 Is a Promissory Note

Intchains: The First Half’s “Revenue” Was a Related-Party Sale, and 2027 Is a Promissory Note

Intchains Group is the small altcoin-mining ASIC design house that trades on the Nasdaq under ICG, sells its machines under the Goldshell brand, and pairs that hardware business with an Ethereum treasury. Its first-half 2026 report, out August 20, should end the argument about whether the mining-hardware business is in trouble. It is not in trouble. It is effectively gone.

The headline number: first-half revenue of RMB11.1 million, roughly US$1.6 million, down 93.7% from RMB175.6 million a year earlier. Strip the detail and it gets worse. Of that RMB11.1 million, RMB10.9 million was a sale of non-core chip inventories to a related party — a transaction between connected entities, not an arm’s-length customer. The third-party revenue for an entire half-year was roughly RMB0.2 million. Call it zero.

The market read the number the way it should. The shares fell about 4% on the session, to roughly $0.84, then dropped about 14.3% in extended trading toward $0.72 once the full filing hit the tape. On the same session, Canaan and Bitdeer — the larger listed crypto-mining-hardware names — rose around 16% and 8%, respectively. Whatever the sector backdrop was doing, the tape treated this report as a company-specific problem, because it was.

Tape-out is not a turnaround

When core revenue rounds to zero, a company needs a story to sell. Intchains’ story is a “next-generation mining ASIC”: tape-out completed in July, engineering samples moving through laboratory validation, commercial launch targeted for Q4 2026, architected for proof-of-work with a focus on performance per watt on a mature process node, aimed at “a specific new cryptocurrency” in a niche where competition is “not intense.”

Be honest about what a tape-out is: it is the moment the finished design is sent to a fab to be manufactured. It is a milestone, not a product. There are no third-party specs, no named coin, no independent benchmarks for the claimed leading performance — none of it is verifiable, and none of it will be until a named coin and an order book exist. The astute investor’s response to “competition is not intense” is to ask why. For an unnamed coin with immature economics, the answer is that the market is too small for anyone with real scale to bother defending.

The detail that should bother investors more is the R&D budget. The company cut research and development spending 46% year over year in the first half, to RMB22.4 million, as part of a restructuring expected to save an estimated RMB23.1 million of annual labor cost. One of the two claims has to be wrong: either this chip is the most important product in company history, or the people who design the chips are the first cost to cut. Management is trying to sell a flagship product and a cost discipline plan at the same time, and the two narratives are in direct contradiction.

Product launches have a habit of not landing

The record says what this company’s flagship launches actually produce. In 2025 alone, it launched an ALEO miner, the Goldshell Byte dual-miner, Dogecoin and XTM machines, plus multiple iterations of each. Full-year 2025 revenue fell 21.6% to RMB220.9 million, the business swung to an operating loss of roughly US$15 million and a net loss of US$7.4 million, and cost of revenue jumped 57% on impairment charges for excess altcoin-miner inventory that nobody bought. Demand that was strong at the start of 2025 turned “softer” in the back half, soft enough that third-quarter revenue was already down 84.8% year over year before 2026 even started. This is not cyclical bad luck. It is a business that repeatedly builds hardware for coins whose mining economics expire faster than the inventory moves.

Management’s credibility problem goes beyond the demand pattern. A February 2026 directive from eight Chinese authorities, including the People’s Bank of China, extended the country’s cryptomining restrictions to offshore providers — an escalation of the 2021 ban that reaches exactly the export-oriented model Intchains claims to run. On the Q4 call in May, management said the notice was not expected to have a material adverse impact. By August, the H1 report admitted the restrictions “added pressure” to sales. That is a management claim falsified within a single reporting cycle, and it is the reason every forward comment from this team should be discounted until the engineering and sales evidence arrives.

2027 is the only year in the record that hasn’t disappointed

Management told the call that 2027 could be the best year among five years of the company. Note the trick: 2027 is the one year in the table with no results attached to it — the only year that cannot disappoint until it arrives. The claim quietly stacks three conditions that must all hold at once: cryptocurrency prices for the legacy coins (Dogecoin, Aleo) recover; the mystery coin’s mining economics hold up; and the new ASIC moves from tape-out through mass production quickly enough to matter. Nothing in this company’s operating history says those three arrive together.

There is also a structural problem specific to small-coin mining that no amount of performance-per-watt cleverness fixes. A proof-of-work miner’s payout is proportional to coin price and machine hashrate, divided across the whole network’s difficulty, minus power and machine cost. For a niche coin, network difficulty is largely set by the machines being shipped — so every unit a hardware vendor sells raises difficulty and lowers the customer’s yield. The TCO arithmetic is why this niche has “not intense” competition. It is not that Intchains found a corner nobody else saw. It is that nobody with real wafer economics wants to own that corner.

The rest of the story is a balance sheet, not a chip company

What the equity actually approximates now is an Ethereum holding company with a chip option attached. The balance sheet carries roughly US$68 million in cash and short-term investments plus another US$14 million of Ethereum by fair value — about 9,176 ETH, with 4,556 ETH staked as of August 20, mostly through its own Goldshell staking platform. Against roughly 60.7 million shares outstanding, the current price implies a market capitalization in the mid-$40s-to-$50 million range. In other words, the stock trades for less than the cash alone, with the token pile thrown in.

So the market already prices the operating chip business at negative value. Given a half-year in which real third-party revenue was on the order of a rounding error, and a management team whose prior “new product” cycles ended in inventory write-downs, negative is not a mispricing — it is a reasonable clearing price for the option. The $15 million share repurchase management authorized, funded from that cash, is the one concrete thing this team can do that does not require the industry to cooperate.

The honest frame: Intchains is now roughly a small Ethereum-yield fund with a promissory note that says “2027.” The judgment separating astute investors from gullible ones is whether that note is treated as an engineering commitment or as a marketing asset. The company’s own launch record, its own R&D cuts, and its own falsified regulatory claim all say the latter. What would change the view: naming the coin and publishing third-party specs before Q4, a buyback that actually executes, and a reporting period in which third-party revenue is more than a rounding error rather than a sale to a related party. Until then, treat this ASIC like every launch that preceded it. Wait for the evidence.