Cango's AI Pivot Hits a Milestone — First Customer Contract Signed, Mining Deliberately Shrinks
From megawatts to customers
In the second quarter of 2026, Cango reported a headline net loss of $81.6 million, but the real story is what the company did after June 30. The CEO, Paul Yu, opened the call with a clear framing: "On the AI side, since the end of the second quarter, we have made real progress on infrastructure and signed our first customer contract, moving that business from build-out into commercialization." That contract — signed post-quarter — marks the moment Cango transitions from a pure Bitcoin miner into a potential AI compute provider with actual revenue. It is a deliberate, company-unique pivot that has been gestating for several quarters, and this quarter provides the first concrete validation.
The AI infrastructure build-out at the Georgia LN site reached completion in early July, with containers arriving and GPUs being installed in phases. Management stressed that none of this is reflected in Q2 results, but they expect to begin recognizing customer contract revenue in the third quarter. When asked about magnitude, CFO Simon Tang said the initial contribution will be "modest," but the strategic importance is disproportionate. This is the first real-world proof that the company can convert its legacy mining power into AI service. As CEO Yu noted: "We will continue to run mining and AI as parallel businesses," and the company is evaluating new sites and even building its own capacity.
The progress is not instantaneous. In the February 2026 call (Q4 2025), the CEO had been cautious, saying "our near-term focus is on validating the commercial models and evaluating unit economics," and in the March 2026 call, the CFO reiterated that revenue would come "in the second half of this year." Now, with the first contract already signed and the Georgia site operating, the validation is nearly complete. The commercial model — a mix of bare-metal GPU hosting and colocation at existing power sites — is not yet fully defined (no colocation contract signed), but the first deal provides a template for scaling.
This pivot is all the more striking against the company's recent history. Cango's mining hashrate is being deliberately wound down: from over 50 exahash per second (EH/s) a year ago to a combined operating figure of just 27.58 EH/s as of June 30 (19.84 self-mining + 7.74 leased). The company is phasing out older S19 machines and introducing a leasing model where the lessee bears direct operating costs. This is an explicit abandonment of the "scale at all costs" mindset that dominated 2025, when the company aggressively built computing power. Now, the priority is economics and cash preservation.
Harvesting cash and hedging Bitcoin
The financial results bear the scars of this strategy. Revenue fell roughly 50% sequentially to $50.8 million, driven by "our proactive reduction in operational hashrate as we continued to selectively phase out older, less efficient S19 series mining machines and temporarily transitioned some capacity to a hosted leasing model." That deliberately lowered top-line revenue, but also slashed cost of revenue from $99.6 million to $51 million. The net loss was mostly non-cash: $42.9 million in impairment losses and $8.5 million in disposals of mining machines. Underlying cash costs improved to $73,313 per Bitcoin, down 5% sequentially.
While mining is shrinking, the company is adding a risk shield. CEO Yu described a new Bitcoin hedging program implemented during Q2: “The purpose of this program is to manage our exposure to Bitcoin price volatility and provide greater predictability to our operating cash flow.” CFO Tang later detailed the mechanism in Q&A, explaining it as a short-term BTC-denominated loan: “The related short-term positions are reflected on our balance sheet and will be reflected as we continue to execute this program in a disciplined manner.” — Ming Yeung Tang, CFO · 2026-08-31 The facility, roughly $8 million, is sized relative to months of mining production, and if Bitcoin prices fall, the company can repay with Bitcoin it mines. This is a significant departure from the earlier "mine-and-hold" treasury strategy, as noted in prior calls. The company now uses derivatives not to speculate but to stabilize cash flow.
Analysts on the call pressed for clarity on the AI timeline and mining cost trajectory. CFO Tang confirmed that Q3 mining hashrate will not change significantly, though summer curtailments may affect production. He also provided a nuanced view on power costs, explaining that many hosting contracts include a price reduction mechanism that automatically lowers power prices when Bitcoin prices fall — a built-in hedge. This mechanism, along with the new BTC loan structure, helps protect margins in a volatile market.
We will continue to evaluate the mix between self-mining and leasing based on economics rather than scale. We will keep phasing out less efficient legacy capacity.
The decision to prioritize cash over growth is a recognition that the mining business is no longer the growth engine. Instead, it becomes a cash cow to be milked selectively while capital migrates to AI. The company still holds 1,056 Bitcoins on its balance sheet, but liquidity is thin: cash plus cryptocurrencies total roughly $23 million against $31.2 million in long-term debt. The company has already raised $10.1 million from shareholders this year and arranged an additional $65 million equity investment, giving it dry powder to fund the AI expansion without taking on debt.
In the context of the broader tape, Cango's pivot stands out. While some large-cap miners with AI ambitions have seen their valuations compress as investors question the CapEx cycle, Cango is taking a deliberately capital-light, modular approach. It is essentially converting its existing power sites into earning assets, one container at a time. The first customer contract, though small in immediate revenue, is an important proof point that the model works — a step many larger operators have not yet achieved. The AI deployment strategy is now backed by operating experience and a real revenue line, not just PowerPoints.
The next few quarters will be critical. If the Georgia site ramps as planned and additional customers sign, Cango could be a rare example of a miner successfully transitioning into an AI compute provider. But if the modest revenue fails to scale, the company could be left with a much smaller mining business and an unproven new venture. For now, the market should watch the Q3 report, where AI revenue is expected to appear, and the evolution of the Bitcoin hedging program, which could provide a template for how miners manage price risk in an increasingly turbulent crypto market.