Good morning. It's Monday, July 20th, 2026. Let's get into it.

**First up — SpaceX pushes Starship Flight 13 to Thursday.** After the automatic abort on July 16, where four of thirty-three Raptor 3 engines on Booster 20 failed to ignite at T-minus zero, SpaceX is now targeting Thursday, July 23rd for the re-attempt. The launch window opens at 2:45 PM Pacific from Starbase. SpaceX is replacing two of the four failed engines and has confirmed the stack is still fully integrated at Pad 2. This comes after the FAA officially closed the Flight 12 mishap investigation on July 13th with four corrective actions — the clearance that made this week's attempt possible. Flight 13 remains the mission that carries the first Starlink V3 satellites, which feature deployable solar arrays and will be tested with a new ground station in South Africa. The booster landing attempt is still on the table — same hardware class that experienced the cascade failure destroying Booster 19 on Flight 12.

**The World Artificial Intelligence Conference wrapped in Shanghai today** — and with it, the formalization of a two-track global AI governance order. President Xi Jinping delivered his first-ever public keynote on AI policy, declaring Beijing's intent to shape international rules for the technology. The headline product: WAICO, the World Artificial Intelligence Cooperation Organization, now has 29 signatory nations — Russia, Brazil, Pakistan, Indonesia, and two dozen others, but not a single Western democracy. This creates a parallel governance track to the EU AI Act and the G7 framework. Huawei also debuted the Atlas 950 Supernode — 1,024 Ascend chips per node, scaling to over 8,000 networked — the world's largest AI server cluster built entirely on domestic silicon. Xi also pledged 5,000 AI training slots for developing nations. The institutional split is locked in, and enterprises operating in both spheres now face incompatible compliance regimes.

**Blackstone is betting on the hardware layer of the robotics boom.** The world's largest alternative asset manager agreed to invest in Futronic, a South Korean supplier of high-precision actuators for industrial robots and automotive systems. The deal values Futronic at about one trillion won — roughly 676 million dollars. This is notable because actuators are the least visible but arguably most essential component in any robot: the joints, the motors, the things that actually make movement possible. As humanoid robot production scales globally, the companies supplying core motion components are becoming the pick-and-shovel plays of physical AI. Blackstone isn't alone — SK Hynix chairman Chey Tae-won noted over the weekend that foreign governments now treat AI memory access as economic security, and customers are demanding 60 to 100 percent more AI memory capacity in 2027.

**Speaking of supply constraints — Samsung and SK Hynix are warning that AI-driven memory shortages could stretch into 2027 and beyond.** HBM production is eating an increasing share of wafer capacity at both companies. TrendForce estimates high-bandwidth memory will account for about 9 percent of total DRAM bit supply this year, rising to 13 percent in 2027. Because HBM requires more advanced manufacturing and consumes more silicon than conventional DRAM, it's effectively displacing standard memory production and keeping prices elevated. SK Group is now even considering building a semiconductor plant in the United States to expand supply. Both Samsung and SK Hynix are also developing a new memory tier — CXL interconnect memory — to handle AI workloads that have outgrown what HBM bandwidth alone can deliver.

**And finally, Morgan Stanley has quietly become Wall Street's top bank for AI infrastructure debt.** The firm has arranged more than 40 billion dollars in AI infrastructure bond offerings this year — surpassing Goldman Sachs. Deals include a 3.2 billion dollar bond for Google-backed data center developer TeraWulf, a 27 billion dollar debt package for Meta's Hyperion data center project with Blue Owl, and advising Broadcom on a 35 billion dollar chip financing deal. LSEG data shows Morgan Stanley pulled in 2.3 billion dollars in capital markets fees in the first half of 2026 alone, up from 1.4 billion a year earlier. The AI buildout isn't just a technology story — it's becoming one of the largest capital formation narratives in modern finance.

That's all for today. Keep building.