Weekly results: miners are massively moving into AI, bitcoin tests support, and hackers are mastering neural networks
The outgoing week proved telling: bitcoin failed to sustain July's momentum and retreated to lows, miners are massively converting capacity into AI infrastructure in search of a new revenue model, and cybercriminals are increasingly using artificial intelligence in their schemes. Let's break down the key events in detail.
Bitcoin: Return to August Lows
The market reversed. On August 14, the leading cryptocurrency broke through the $63,000 level, erasing all the gains of the previous seven-day period, which had ended at $65,200. Notably, early in the week, CryptoQuant analysts pointed to resistance in the $67,000 and $72,000 zones, while BlackRock spoke of a shift in investor sentiment. However, reality turned out more prosaic: the market, according to Glassnode, found itself in a "compressed" state, squeezed between the median realized price of $63,000 and the cost basis of short-term holders at $68,700.
Breaking the lower boundary of this range is a negative signal. Glassnode warns that in such a scenario, the path is open to $58,500. At the time of writing this review, bitcoin is consolidating near $63,000, losing 3.3% over the week. Ether fell 2.1% to $1,880, and among the top cryptocurrencies, only Hyperliquid's token is in the green (+4.7%). The Fear and Greed Index is frozen at 34, corresponding to the fear zone, while the total market capitalization shrank from $2.22 trillion to $2.17 trillion.
Institutional demand also weakened: spot bitcoin ETFs recorded their largest weekly outflow since early July at $398.7 million. Ether funds lost $2.2 million after an inflow of $244.9 million the previous week.
Miners: A Shift in Business Model
A tectonic shift is underway in the mining industry. Miners selling bitcoin is no longer a crisis measure—it is a deliberate strategy to finance the transition to AI. A striking example is former Bitfarms, now Keel Infrastructure, which fully decommissioned its mining sites in the U.S., preparing them for data centers for high-performance computing.
Riot Platforms is acting even more decisively. The company signed a 20-year capacity lease contract with a leading AI lab worth $9.1 billion, sold 4,300 BTC, and raised up to $573 million for building an AI campus in Texas. The largest U.S. miner, MARA, sold 23,093 BTC for $1.6 billion in the first half of the year, citing the financing of growth. Hyperscale Data also joined the list, selling 685 BTC for $43 million.
The economics of mining are indeed pushing toward such a pivot: fee revenues have fallen to a ten-year low, and the hashrate of public companies has dropped 21.2% over three quarters. This is not temporary market conditions but a fundamental transformation of the industry.
AI in Service of Cybersecurity and Attacks
Artificial intelligence has become a tool on both sides of the barricades. Tellingly, volunteers of Bitcoin Red Team faced restricted access to OpenAI and were forced to return to Chinese models. Meanwhile, attackers face no such issues: the North Korea-linked group Kimsuky is actively using local AI systems to target crypto companies, and Taiwan has revealed details of a hack on government institutions using AI agents.
The founders of the bitcoin service Boltz handed the project over to a group of veterans after a series of attacks allegedly carried out using neural networks. At the same time, the July campaign against Coldcard hardware wallets has ended: according to Galaxy Research estimates, attackers stole at least 1,778.84 BTC ($112.7 million), and no new incidents have been recorded since August 6. Separately, we note data leaks: Trezor reported the compromise of information for 13,689 users through a breach of a logistics partner, and SafePal—40,000 affected.
Regulatory Pressure in Russia
Moscow law enforcement conducted mass searches at "Gorbushka" in a case involving crypto exchangers. This coincided with stricter banking compliance: major banks began requiring corporate clients to confirm the inclusion of counterparties in the non-existent registry of digital currency exchange operators of the Central Bank. In my assessment, this is the logic of banks self-insuring rather than a direct instruction from the regulator, but the uncertainty will not last long—the "Digital Currency" law takes effect on September 1.
My comment: the miners' shift to AI is not panic but a rational response to structural changes in the mining economy. However, for the market, this means additional downward pressure on bitcoin's price in the medium term until new business models stabilize.