D

$MU Bargain of the Century PE Ratio: 15.5 Sales Ratio: 2.33 50% Increase in HBM (AI Memory) Sequentially. DRAM/NAND prices are surging.

$MU Nine Price Target Upgrades 🤯 Barclays went from $450 to $675 (Street High). The reasoning is the same everywhere: HBM4, Durable AI demand, and long-term agreements. Goldman Sachs is the lone Hold, and even they raised their target. The wall of skepticism is crumbling

Sunbathers in Dolores Park are taking advantage of the heat wave in San Francisco on Saturday. 📷: Aidin Vaziri/S.F. Chronicle

Can this be a solution?

STARTUP TEASES BITCOIN MINING IN SPACE US startup Starcloud says it plans to launch a dedicated bitcoin mining rig into orbit later this year, potentially becoming the first company to mine bitcoin in space. The announcement came from Starcloud CEO Philip Johnston during an appearance on the HyperChange YouTube channel. Johnston said the company is currently focused on building its orbital data center constellation but confirmed that bitcoin mining hardware is also part of its near term plans. “There’s also bitcoin mining,” Johnston said. “We’ll have some bitcoin mining ASICs on the second spacecraft launching later this year.” The move follows Starcloud’s successful test of running an Nvidia H100 GPU in orbit, demonstrating that high performance compute infrastructure can operate in space. Johnston says the long term vision centers on energy. Orbital data centers and mining rigs could tap continuous solar power in space, potentially providing a cheaper and more scalable energy source than what is available on Earth. Compared with enterprise AI hardware like Nvidia GPUs, which can cost $30,000 or more, bitcoin mining machines are relatively inexpensive, typically ranging from $600 to a few thousand dollars. “The cat is out of the bag,” Johnston said on X. “Starcloud will be the first to mine bitcoin in space.” He added that the opportunity could be massive, noting that bitcoin mining currently consumes roughly 20 GW of power globally and arguing that in the long run, a significant portion of that compute could move off planet.

$MU Thanks for 5-year partnership with NVIDIA, Micron extends its leadership in Low-Power DRAM for AI Servers (SOCAMM2) which is emerging as one of the most critical memory innovations for scaling next-generation AI infrastructure efficiently. SOCAMM2 addresses the core energy bottleneck head-on. In the era of energy-constrained data centers, SOCAMM2's importance lies in its superior power efficiency. It achieves over 20% better power savings than Micron's previous LPDDR5X generation and more than two-thirds reduction compared to equivalent RDIMMs, despite being one-third the size. Data centers face severe power limitations, AI clusters can consume massive energy, with racks potentially exceeding 50 terabytes of CPU-attached memory, making low-power solutions essential to maximize compute density without overwhelming electrical infrastructure or cooling systems. The module's design also supports liquid cooling and easier serviceability, reducing total cost of ownership by allowing quick replacements for degraded memory, which is common in high-utilization AI environments. Micron has pioneered this by delivering the industry's highest-capacity variant, a 192GB SOCAMM2 module using its advanced 1-gamma DRAM process, which provides 50% more capacity in the same compact 14x90mm footprint compared to prior low-power DRAM designs. It is already being sampled to customers, including for Nvidia's next-generation AI platforms like the Rubin architecture. Bullish 🔥