Data: Total value locked (TVL) in DeFi fell by 39% cumulatively in 2026, with TRON and Hyperliquid growing against the trend.

Micron Technology (MU.O) is set to release its fiscal third-quarter results after U.S. markets close on Wednesday. Before the earnings release, Micron’s stock has surged more than 7x over the past year, pushing its market capitalization past $1 trillion. The market expects Micron’s stock to swing roughly 14% post-earnings, corresponding to a market cap shift of over $150 billion. This rally is primarily driven by a severe memory chip shortage fueled by AI demand, which has lifted memory stocks including Micron, Samsung, and SK Hynix. The fiscal third quarter ended in May; consensus estimates call for Micron’s revenue to jump 279% year-over-year to $35.3 billion, with adjusted earnings per share (EPS) rising from $1.91 to $20.28. Given the company’s rapid growth, investors’ focus may extend beyond the just-completed quarter’s performance to its fiscal fourth-quarter guidance. Analysts currently forecast Micron’s fiscal fourth-quarter revenue at $42.5 billion, with adjusted EPS of $24.80. If the company delivers another set of guidance that significantly beats expectations, its stock could continue rising; however, at current valuation levels, any miss on key metrics could trigger a sharp pullback. Profit margins are also a core variable. Micron’s fiscal second-quarter gross margin hit 74.4%, up from 36.8% in the year-ago period and 56% in the first quarter. The company previously projected a gross margin of 81% for the third quarter. If gross margins continue to rise in the fourth quarter, it would signal that the memory pricing environment remains tight. Additionally, the market will watch whether management maintains its view that "supply tightness for data center and AI-related memory will persist beyond 2026". Micron previously stated that for some key clients, it can only meet 50% to two-thirds of their demand in the medium term.

Capital.com analyst Daniela Hathorn stated in a report that upcoming U.S. economic data will determine whether the current rally of the U.S. dollar has further upside potential. She explained that if inflation and labor market data continue to support the narrative that interest rates will remain elevated for an extended period, the dollar should continue to draw support. If price pressures start easing faster, markets may begin to question whether current interest rate expectations have become overly aggressive. The PCE data (the U.S. Federal Reserve’s preferred inflation gauge) and initial jobless claims data are set to be released on Thursday, while the non-farm payrolls report is due on July 2. (Jin10)

On Wednesday, multiple Wall Street institutions issued key rating adjustments covering firms including Nvidia, Tesla, AMD, Reddit, and Arm. Seaport reaffirmed its "Sell" rating on Nvidia, citing growing burdens on and off Nvidia’s balance sheet, plus Broadcom’s entry into the cycle financing trend, as signs of rising financing constraints facing the AI industry. Stifel initiated coverage on BlackBerry for the first time, assigning it a "Buy" rating, noting the market still misdefines BlackBerry; the company is no longer just an automotive supplier, but a critical software layer in the physical AI stack, and partners with chip firms including Nvidia, Qualcomm, and AMD. UBS reaffirmed its "Buy" rating on AMD and raised its target price for the chipmaker from $470 to $670, stating it is more positive on AMD as its standalone CPU racks gain market recognition. TD Cowen reaffirmed its "Buy" rating on Arm and lifted its target price from $265 to $475, noting Arm’s 2031 fiscal year target of $15 billion for AGI CPUs is reasonable, though GPU-CPU pairing rates and per-core pricing remain key variables. RBC reaffirmed its "Outperform" rating on Tesla, forecasting the automaker’s second-quarter deliveries will hit 405,000 units, above the Visible Alpha consensus estimate of 401,000. Needham reaffirmed its "Buy" rating on Reddit, saying the platform holds a rare competitive position largely insulated from the generative AI storm. Separately, Citizens initiated coverage on Bitdeer, CleanSpark, Keel Infrastructure, and MARA Holdings, assigning all an "Outperform" rating, noting these bitcoin miners are repurposing their existing power capacity to deliver high-performance computing to hyperscale clients and large enterprises.

According to Bitget data, spot silver plunged 4.00% intraday, currently trading at $59.06 per ounce.

The U.S. Department of Justice (DOJ) announced Tuesday that it has seized cloud computing accounts belonging to a subsidiary of Cambodia-based Huiwang. The DOJ labeled Huiwang as “one of the world’s most prolific criminal marketplaces,” accusing it of facilitating the laundering of billions of dollars in proceeds from crypto investment fraud and cyber scams. The agency said the accounts provided technical support for Huiwang’s operations, enabling criminals to transfer, move and hide funds, and convert funds into the banking system undetected. Assistant Attorney General A. Tysen Duva of the DOJ’s Criminal Division noted that the seizure targets a technical backbone that allowed billions in fraudulent proceeds to be moved and hidden, with a large portion of the funds coming from scam hubs in Southeast Asia. Court documents show the accounts helped operate Huiwang Guarantee, a Telegram-based marketplace where sellers trade stolen bank card and identity data, malware proceeds, and money laundering services for romance scams and investment fraud. The platform also offers escrow services to help criminals, including money launderers, conduct cryptocurrency transactions.

Qualcomm announced it has reached an agreement to acquire Modular, aiming to strengthen its software foundation for generative AI and agentic AI in data center and edge environments. The transaction is expected to close in the second half of 2026, pending customary closing conditions and relevant regulatory approvals. Qualcomm stated that Modular offers an open, AI-native software stack that enables AI to run efficiently across diverse hardware architectures. Its unified platform supports CPU, GPU, NPU, and custom ASIC architectures, eliminating the need to rewrite models for each type of accelerator. For developers and enterprises, this means they can build once and deploy across various environments while reducing total cost of ownership. Qualcomm noted that as AI scales, efficiency rather than capability is emerging as the limiting factor. Performance per watt impacts inference costs, and these costs determine whether AI can scale. This acquisition will further enable Qualcomm to deliver a chip-agnostic compute layer across devices, edge, and data centers, boosting performance per watt, increasing hardware flexibility, and expanding its open developer ecosystem.