$HOOD Morgan Stanley's HOOD Summit preview names U.S. perpetual futures as the product most likely to be announced in Houston next week, and it is not in the bank's model. Robinhood already runs perps overseas, on chain and through Bitstamp, where the offering has grown to include commodity perps. On the Q2 call Vlad Tenev said of the CFTC, "we've been in conversations with CFTC, and they've been constructive." Morgan Stanley's point is that because the product already exists internationally, the technical work for a U.S. version is small. Tenev would not give a date and said product timelines are what events are for. The SEC's Innovation Exemption on Sept 17 opened a path for tokenized U.S.-listed stocks to trade on chain through qualifying venues, with a five-year sunset and limits on symbols and volume. Tokens have to carry the same rights as the underlying shares. Robinhood's current Stock Tokens are an international product that gives price exposure without those rights, so a U.S. version needs a different structure. With twelve days between the SEC action and the Summit, Morgan Stanley expects a roadmap update rather than a live product. The bank puts Stock Token balances above $150 million and Robinhood Chain TVL above $1.5 billion. Neither U.S. perps nor U.S. tokenized stocks are in its forecasts. Overweight, $150 target, against $124.25 at the Sept 22 close.

Why one satellite company needs 3 spacecraft and another needs 66. $RKLB, $IRDM and $VSAT are the examples. A geostationary satellite orbits at the same speed the Earth turns, so from the ground it never moves and one dish points at it once. To do that it has to sit far enough out for one orbit to take a day, and every signal makes that long trip up and back. A low Earth orbit satellite flies close and fast, crossing the sky in minutes. That cuts the delay and lets a handheld talk to it, but none of them stays overhead, so you need a fleet and a steady supply of replacements as the low ones wear out. $IRDM 's numbers: it did $871.7M of revenue in 2025 with operational EBITDA of $495M, a 57% margin, and $RKLB is paying $54 a share for it, half in cash, with closing expected mid-2027. @Peter_J_Beck , Founder and CEO of Rocket Lab: "By marrying Iridium's deep heritage, trusted infrastructure, and highly sought-after spectrum with Rocket Lab's extensive and proven launch and manufacturing capabilities, we have the capability to unlock entirely new markets." A launch company buying a LEO operator is buying its own recurring customer. A GEO operator never needs it that often. Find the altitude first. It tells you the delay, the antenna, the fleet size and how often the launch bill comes due. Which orbit does the satellite stock you follow fly in? It's not a sprint, it's a marathon.

$CIFR 𝗕𝗮𝗿𝗯𝗲𝗿 𝗟𝗮𝗸𝗲 𝘄𝗲𝗻𝘁 𝗳𝗿𝗼𝗺 𝘁𝘄𝗼 𝗯𝗶𝗴 𝗵𝗮𝗻𝗱𝗼𝘃𝗲𝗿𝘀 𝘁𝗼 𝗮 𝗵𝗮𝗹𝗹-𝗯𝘆-𝗵𝗮𝗹𝗹 𝗿𝗮𝗺𝗽. Under the 2025 terms: 1. 168 MW delivered to Fluidstack by September 2026 2. Rent from October 2026 3. Another 39 MW by January 2027 Under today's amendment: 1. Delivery by individual data hall, Q4 2026 through Q1 2027 2. Rent starts on each hall as it is delivered 3. First rent in Q4 2026 $CIFR cited change orders and evolving tenant requirements, and says it is on track with the revised schedule. The tenant is already inside. $CIFR 's Q2 update described partial occupancy and network racks going in. That is early access for setup. Rent is tied to a finished hall that is powered, cooled and handed over. Black Pearl already works this way. $CIFR delivered first capacity there in August, two months early, with rent started while the other halls were still in fit-out. My estimate on the rent math: 1. About 10 halls of roughly 20 MW, since 168 MW splits into 8 x 21 MW and 39 MW into 2 x 19.5 MW 2. ~$383M a year at full run-rate from the $3.0B and ~$830M contracts, about $3.1M a month per hall 3. An even ramp puts ~$48M less rent into Q4 and Q1 than the old plan 4. Each month one hall slips moves about $3.1M In exchange: 1. 10 more years with a leading AI lab, ~$5.2B added, over $9B contracted 2. $CIFR carries the first $359.3M of costs above the original budget 3. Above that, the tenant repays half over 20 years as extra rent at a contracted return Rack power for the next GPU generation is climbing fast, which would be one reason to rework halls mid-build. That one is my guess, not something $CIFR said. I hold $CIFR.

They say they tested some $IREN capacity themselves — fine. “We’ve heard the stories” + unnamed users for the worst-site claim is not enough. Without disclosure this is just paid actor. I am not buying this fake news.

$IRDM 𝘀𝘁𝗼𝗰𝗸𝗵𝗼𝗹𝗱𝗲𝗿𝘀 𝗮𝗽𝗽𝗿𝗼𝘃𝗲𝗱 𝘁𝗵𝗲 $RKLB 𝗺𝗲𝗿𝗴𝗲𝗿 𝗼𝗻 Sept 24. 🚀🚀🚀 1. 99.6% of votes cast in favor 2. 81% of outstanding shares voted yes Each $IRDM share gets: 1. $27.00 in cash 2. $27.00 in $RKLB stock 3. $54.00 notional The stock piece floats on a 10-day VWAP inside a $67.50 to $112.50 collar. Below $67.50 the ratio locks at 0.40 shares, so every $1 of $RKLB weakness under the floor takes $0.40 off the package. $RKLB closed at $73.61 on vote day, about 9% above the floor. $IRDM closed at $48.89, a $5.11 spread to $54, about 10.5%. Since the June announcement: 1. HSR lapsed Aug 12 2. S-4 went effective Aug 26 3. $1.94B ATM completed at ~$66.38, $3.6B bridge terminated 4. Iridium's $1.775B term loan amended to stay in place after close "Today's vote is an important milestone in bringing together Rocket Lab and Iridium to create a next generation space powerhouse," said Sir Peter Beck, Rocket Lab Founder and CEO. The FCC license transfer, filed Aug 10, is still pending. Close is expected by mid-2027.

$IRDM 𝘀𝘁𝗼𝗰𝗸𝗵𝗼𝗹𝗱𝗲𝗿𝘀 𝗮𝗽𝗽𝗿𝗼𝘃𝗲𝗱 𝘁𝗵𝗲 $RKLB 𝗺𝗲𝗿𝗴𝗲𝗿 𝗼𝗻 Sept 24. 🚀🚀🚀 1. 99.6% of votes cast in favor 2. 81% of outstanding shares voted yes Each $IRDM share gets: 1. $27.00 in cash 2. $27.00 in $RKLB stock 3. $54.00 notional The stock piece floats on a 10-day VWAP inside a $67.50 to $112.50 collar. Below $67.50 the ratio locks at 0.40 shares, so every $1 of $RKLB weakness under the floor takes $0.40 off the package. $RKLB closed at $73.61 on vote day, about 9% above the floor. $IRDM closed at $48.89, a $5.11 spread to $54, about 10.5%. Since the June announcement: 1. HSR lapsed Aug 12 2. S-4 went effective Aug 26 3. $1.94B ATM completed at ~$66.38, $3.6B bridge terminated 4. Iridium's $1.775B term loan amended to stay in place after close "Today's vote is an important milestone in bringing together Rocket Lab and Iridium to create a next generation space powerhouse," said Sir Peter Beck, Rocket Lab Founder and CEO. The FCC license transfer, filed Aug 10, is still pending. Close is expected by mid-2027.

$APLD's Polaris Forge 1 in Ellendale, North Dakota won Project of the Year, small population category, from the Mid-America Economic Development Council. "Applied Digital's investment represents a turning point for our community, bringing new energy, new jobs and new opportunities," said Ellendale Mayor Don Flaherty. The campus has 175 MW live after the first 75 MW phase of Building 2 came online June 30. All 400 MW is leased to $CRWV on roughly 15-year terms, about $11B of contracted revenue, and the site is engineered to scale to 1 GW over time.

🚨🚨🚨🚨🚨 Morgan Stanley modeled $CIFR Barber Lake 8 days ago as a 10-year, $3.8B lease delivering in Q3 2026. Today $CIFR made it 20 years and over $9B, with a leading AI lab committed to years 11 through 20. Delivery also moved. Data halls now phase in from Q4 2026 through Q1 2027, with first rent in Q4. On Sept 11 I asked whether Barber Lake would get its own release or wait for the Q3 update. It got its own release today. $CIFR took the lease from 10 years to 20, and contracted revenue from $3.8B to over $9B. Delivery moved from September to phases running Q4 2026 through Q1 2027. First rent is still expected in Q4.

$SMCI is now shipping $NVDA Vera Rubin NVL72 racks, integrated with its Data Center Building Block Solutions and the DLC-2 direct liquid cooling stack. Per rack: 1. 72 $NVDA Rubin GPUs and 36 Vera CPUs, operating as one machine 2. 18 1U compute trays, each with 4 Rubin GPUs and 2 Vera CPUs 3. 9 sixth generation NVLink switch trays, 216 TB/s of scale-up bandwidth 4. 20.7 TB of HBM4, up to 54 TB of LPDDR5X 𝗢𝗻𝗲 𝗦𝗰𝗮𝗹𝗮𝗯𝗹𝗲 𝗨𝗻𝗶𝘁 𝗶𝘀 1,152 𝗥𝘂𝗯𝗶𝗻 𝗚𝗣𝗨𝘀 𝗮𝗻𝗱 331 𝗧𝗕 𝗼𝗳 𝗛𝗕𝗠4 𝗮𝗰𝗿𝗼𝘀𝘀 16 𝗰𝗼𝗺𝗽𝘂𝘁𝗲 𝗿𝗮𝗰𝗸𝘀. The blueprints size a balanced bill of materials from a 5 MW envelope up to gigawatt scale. Cooling is what Supermicro is really selling, and it comes from its own portfolio: 1. Cold plates, manifolds, hose kits 2. In-row CDUs rated 1.8 MW each, N+1 redundancy 3. Rear door heat exchangers for residual heat 4. Facility-side cooling towers Every rack is tested with the full stack before it leaves. "We have spent years building the liquid-cooling stack, the manufacturing capacity, and the deployment teams for exactly this moment," said Charles Liang, president and CEO, Supermicro. Supermicro also said on its own X account that the first systems went to SpaceX AI, built at the expanded San Jose facility. Shipping first is the win. The open question for $SMCI is the margin these racks carry, because this is a company that has taken thin ones before to hold volume.

𝗠𝗼𝗿𝗴𝗮𝗻 𝗦𝘁𝗮𝗻𝗹𝗲𝘆 𝗰𝗼𝘂𝗻𝘁𝗲𝗱 71 𝗺𝗲𝗻𝘁𝗶𝗼𝗻𝘀 𝗼𝗳 $TSLA 𝗶𝗻 𝘁𝗵𝗲 $SPCX 𝗦-1. Adam Jonas's Sept 20 note uses them to map where the two Musk companies already lean on each other. The bank thinks the next leg of AI is physical: robots, robotaxis, and the power and networks behind them. Most of the link is already on the books. $SPCX buys Megapacks to run Colossus. $TSLA holds $SPCX stock from the xAI merger. The two are building Terafab with Intel in Grimes County, Texas. Starlink is going into $TSLA vehicles, starting with Cybercab. "There's no question that there's synergies between Tesla and SpaceX in our futures, definitely, there's a convergence of a kind of what we're all trying to accomplish in the future," said Gwynne Shotwell, $SPCX President and COO, in a CNBC interview ahead of the IPO. MS also argues neither company adds much to the pure frontier LLM market. It sees the open lane as AI in the physical world, and thinks the two chase it separately or, more likely, together. The same note flags the cost of that closeness. Continued co-investment in Terafab and Solarfab could raise governance questions as $TSLA 's share of capex nears the company's operating cash flow. MS keeps $SPCX at Overweight with a $300 target and $TSLA at Equal-weight with a $400 target.

Transparency trade update.🚨🚨🚨 Bought the $TEM Jan 21 2028 $55 calls on May 15 when we were still in the trenches. $13.80 cost basis. Long-dated LEAPs only. Sold a portion Thursday at $40 when it hit $80. Roughly 190% realized. What’s left is marked at $37.45, still +171% unrealized. Additionally, haven’t sold a single share. X does not dictate my trades. I take chips off when the numbers work for me. The rest stays. The original write-up still stands. Pricing step-ups, Personalis, vaccine sequencing, data licensing. None of that changed because the stock ran 30% in a week. This is still the start. It’s not a sprint, it’s a marathon. -PSI

𝗧𝗲𝗺𝗽𝘂𝘀 $TEM 𝘀𝗮𝘆𝘀 𝗶𝘁𝘀 𝗼𝘄𝗻 𝗳𝗼𝗿𝗲𝗰𝗮𝘀𝘁 𝗳𝗼𝗿 𝗣𝗲𝗿𝘀𝗼𝗻𝗮𝗹𝗶𝘀 $PSNL 𝗶𝘀 𝘁𝗵𝗲 𝗹𝗼𝘄𝗲𝗿 𝗼𝗳 𝘁𝗵𝗲 𝘁𝘄𝗼 𝗶𝗻 𝘁𝗵𝗲 𝗳𝗶𝗹𝗶𝗻𝗴. Morgan Stanley hosted Eric Lefkofsky at its healthcare conference on September 15 and published the takeaways that evening. Tempus was content running as Personalis' exclusive distributor and might not have bought the company until early 2027. MolDX kept adding Medicare coverage for NeXT Personal, four indications by May, and a competing bidder appeared, so the timeline moved up. The S-4 carries a Personalis forecast and a Tempus forecast for the same business, and they do not match. Management told Morgan Stanley the gap comes mostly from how many indications each side assumes get covered and what each test gets paid, and that the Tempus set is the lower one. Personalis has closed above the $16.25 offer on and off since the vaccine readout in August. Lefkofsky's position, as the note relays it, is that the competing bid already played out and that Tempus and Merck between them hold about a quarter of the vote and want the deal closed. From the note: "Management also noted that PSNL trading above the deal price does not make sense in the context of competitive bids having already played out." Personalis holds a termination right of its own if Tempus stock sits under the lower floor when the exchange ratio is set, and the S-4 says Tempus has closed below that line on several days since signing.

🚨🚨🚨🚨🚨 $CIFR has extended Barber Lake from a 10-year lease to 20 years. 𝗖𝗼𝗻𝘁𝗿𝗮𝗰𝘁𝗲𝗱 𝗿𝗲𝘃𝗲𝗻𝘂𝗲 𝗴𝗼𝗲𝘀 𝗳𝗿𝗼𝗺 $3.8𝗕 𝘁𝗼 𝗼𝘃𝗲𝗿 $9𝗕. 1. ~$5.2B added by a leading AI lab 2. Binding 10-year commitment that starts after the Fluidstack lease ends 3. Separate lease, economics substantially consistent with the current one The original Fluidstack deal was signed one year ago today. The second decade was two five-year options then. It is now contracted, with a second tenant lined up behind the first. Tyler Page, CEO: "Barber Lake was designed as a long-lived, mission-critical asset." The delivery schedule was revised with the amendment. 1. Data halls deliver in phases, Q4 2026 through Q1 2027 2. Rent starts hall by hall, first rent in Q4 2026 3. $CIFR says it is on track Change orders pushed costs above the original budget. 1. $CIFR covers the first $359.3M 2. Above that, the tenant repays 50% over 20 years as rent, at a contracted return 3. The other 50% stays with $CIFR