Newsletter · · Ashutosh Agarwal

AST SpaceMobile Clears FCC on Grain Spectrum as T-Mobile Signal Builds - The Satellite & Space-Comms Race - Week of August 21, 2026

The Satellite & Space-Comms Race for the week of August 21, 2026: the FCC cleared AST SpaceMobile to test Grain Management's low-band spectrum under a clause requiring compliance with existing AST and T-Mobile agreements, Rocket Lab posted a record quarter and a near-billion-dollar defense year, and SpaceX rallied roughly 40% as its first lockup passed and the story hardened into an AI-infrastructure bet.

The Satellite & Space-Comms Race

Week of August 14–21, 2026: AST SpaceMobile Clears FCC on Grain Spectrum as T-Mobile Signal Builds

If last week belonged to the earnings prints, SpaceX's first-ever quarter, AST's Q2, Rocket Lab's $8B swing for Iridium, this week belonged to the plumbing underneath them. The FCC quietly handed AST SpaceMobile the keys to test a chunk of nationwide low-band spectrum that traces straight back to T-Mobile. Rocket Lab's defense haul for the year crossed nearly a billion dollars. And SpaceX, three weeks after a bruising post-earnings drop, ripped roughly 40% higher as its first big share-lockup came and went without the flood of selling everyone feared.

The through-line: this is no longer a story about whether satellites can talk to ordinary phones. That's largely settled. It's now a story about who controls the spectrum, who gets the carrier contracts, and, increasingly and strangely, how much of all this is really an artificial-intelligence bet wearing a rocket costume.

A quick housekeeping note on sourcing, because it matters this week: almost none of what follows is management talking on the record. It's investors and commentators reading tea leaves from FCC filings, LinkedIn hires, and earnings decks. We flag who's who throughout, investor, pundit, or (rarely) operator/insider, so you can weight it accordingly.

TL;DR

  • AST SpaceMobile (ASTS): The FCC granted a temporary license for AST to test on Grain Management's 800 MHz nationwide spectrum, and the approval explicitly requires compliance with "existing agreements" between AST and T-Mobile, the first paper-trail hint that the last big US carrier is quietly in the tent. A senior dealmaker hire and Chris Sambar's jump to T-Mobile add fuel.
  • Rocket Lab (RKLB): Record Q2 (revenue $234M, +62% YoY), backlog $2.36B, and a fresh $397M Space Force contract that pushes 2026 defense wins to ~$943M. The $8B Iridium deal is being framed as a cheap way to buy spectrum + a cash-flowing constellation. Neutron's first flight is still officially "end of 2026," and still nobody outside the company believes it.
  • SpaceX (private): Up ~40% off its post-earnings lows as the first staggered lockup passed without a sell-off. The debate has curdled into "is this a rocket company or a leveraged AI-infrastructure bet?" Scott Galloway calls it a "$10 to $30 stock" meme; family offices and endowments keep buying.

What's new

1. AST gets the FCC's green light on Grain's spectrum, and a T-Mobile fingerprint shows up

Source: AST SpaceMobile Podcast, "Why AST SpaceMobile's Grain Spectrum Deal Is Huge" (Aug 18) · AST SpaceMobile Podcast, "J-LEO Confirmed, T-Mobile Next?" (Aug 17) · both investor commentary from the show's retail-analyst host

Here's the plain-English version. Spectrum is the invisible radio real estate a phone uses to connect. "Low-band" spectrum (the 700–850 MHz range) is the good stuff for satellite-to-phone service because it travels far and punches through walls. On Thursday, the FCC granted AST a Special Temporary Authority, a short-term test license, to broadcast on Grain Management's 800 MHz spectrum, described as a 7 MHz-by-7 MHz nationwide low-band block. Grain bought that spectrum from T-Mobile; T-Mobile kept economic upside depending on how it's eventually used.

Why it moves numbers: two reasons.

First, the buildout math. When the FCC approved the sale to Grain, it attached "use it or lose it" milestones, and, crucially, said Grain can satisfy them with a satellite network instead of building thousands of physical cell towers. The host's read of the milestones: roughly 700 kilobits per second within 3 years, 1.4 megabits within 5 years, and 2.1 megabits within 10 years, each across ~90% of the coverage area. Building towers to hit that is slow and hugely expensive; pointing AST's satellites at it is not. That makes AST the obvious partner, and Grain has to name a direct-to-device operator by around November.

"It's clear that AST has been talking, they are talking, to T-Mobile." (the show's host, reading the FCC filing)

Second, the T-Mobile tell. The test authority requires AST's testing to "comply with existing agreements between AST and T-Mobile." That's not a commercial contract (we'd have heard about that), but it's the first hard evidence of some working relationship with the one major US carrier that's been publicly tied to Starlink instead. Stack that against the fact that T-Mobile's CFO spent an Evercore conference in June picking apart Starlink's technical limits, unusual, pointed, and, in the host's telling, the "first crack", and the direction of travel looks clear.

The operator/insider signal worth watching, though it's inferred, not announced: Chris Sambar, the former AT&T network chief who sits on AST's board and championed the original AT&T deal, just joined T-Mobile as chief enterprise officer. You don't move AST's most senior carrier ally into a top T-Mobile seat if the two aren't talking. The host's blunt framing: when the market learns AST has all three US carriers plus a fourth party, "what does the stock do?"

2. Rocket Lab: record quarter, a near-billion-dollar defense year, and Iridium reframed as a bargain

Source: The 7investing Podcast, "Rocket Lab (RKLB) Q2 Earnings & Iridium Deal Explained" (Aug 17), investor (Simon Erickson & Heather Horton) · Astronomy Daily, "Tiny Robots, Alien Caves" (Aug 17), pundit, relaying an operator quote

Rocket Lab's Q2 (reported Aug 10) was a record: $234M in revenue, up 62% year over year, with $581M of new space-systems bookings, $437M of new launch bookings, and total backlog of $2.36B (more than $800M booked in the quarter alone). It was a light launch quarter, only five Electron flights, with the growth engine sitting, as usual, in the satellite-building segment rather than the rockets people actually watch fly.

The bigger strategic news is how 7investing framed the $8B Iridium acquisition (50% stock, 50% cash). Iridium did ~$871M of revenue in FY2025 at roughly a 57% cash-flow margin off 66 operational satellites, more revenue than Rocket Lab's entire launch-plus-manufacturing business. Erickson's point: Rocket Lab isn't really buying a phone company, it's buying ready-to-use spectrum and an in-orbit constellation to jump-start its long-promised "space applications" business (think pay-as-you-go infrastructure in orbit, like renting cloud computing instead of building your own data center). He now expects that revenue line to start around 2028 instead of 2031, and sees backlog more than doubling to $5B+ by early 2027. His early read on the price: attractive, a rare "bolt-on that's ready to go."

On the defense side, the number that matters: a fresh $397M Space Force contract for "Flat-a-lites", flat-panel satellites that stack efficiently in a rocket fairing, for the Space-Based Airborne Moving Target Indicator (SBAMTI) program, tracking airborne threats from orbit. CEO Peter Beck (operator): the company is "proud to contribute to the deployment of a resilient space-based sensing layer that will enhance the joint forces' ability to operate in contested airspace." Add the year's other awards ($190M + $90M + $266M) and Rocket Lab has locked in ~$943M of defense funding in 2026 alone, its largest defense year ever, and the clearest sign it's becoming a real second source to the primes, not a small-cap nipping at SpaceX's heels.

The catch, flagged both by 7investing and by Wall Street Wildlife (Aug 16) (investor): the Flat-a-lites are supposed to fly on Neutron, Rocket Lab's larger, still-unflown rocket. Neutron has slipped from mid-2025 → late 2025 → its current "Q4 2026" target, and there's still no confirmed first-flight date. Seven customers have reportedly signed confidential Neutron launch contracts anyway. As Astronomy Daily put it, that's "either a huge vote of confidence or a slightly nerve-wracking amount of manifest riding on a rocket that hasn't flown yet."

3. SpaceX rips 40% as the lockup dud lands, and the story quietly becomes an AI story

Source: Market Maker, "SpaceX's 40% Rally Explained" (Aug 17), investor · The Prof G Pod, "Is SpaceX Overvalued?" (Aug 17), investor/pundit (Scott Galloway)

Recall the setup: SpaceX went public in June at roughly $135/share and a ~$1.8 trillion valuation, the largest IPO ever, with an unusually tiny 4.9% free float. It then ran to ~$225, cratered ~45% after its first earnings, and drifted toward $100 as its first "staggered" lockup approached (staggered = insiders' shares are released in tranches over time, not all at once at day 180).

Everyone braced for a wave of insider selling. It didn't come. Market Maker's clean explanation of the ~40% rally:

  • The lockup released 911.5 million Class A shares, lifting the float from 4.9% to 11.8%, but institutions who are still bullish simply didn't sell at ~$100–$120, which the market read as a confidence signal.
  • Retail bought the dip hard around $100.
  • Index funds had to buy: as SpaceX's free float rose, Nasdaq-100 trackers needed more shares to match its index weight.
  • A short squeeze: with a reported ~34% of the tradable float sold short, the bounce forced shorts to buy back, adding fuel.

The next real overhang isn't until June 2027, "Elon day", when Musk himself can start selling and the float could climb toward 99%.

But the more important shift is narrative. Two weeks after the IPO, SpaceX borrowed $25B in the year's biggest investment-grade bond deal (debt maturing out to 2056) while sitting on ~$100B of cash, and told investors future funding will come from borrowing, not stock. Galloway's take (bearish, and worth reading in full below) is that the bond deal "clarified the version of the company people are actually buying: not a rocket company… an AI infrastructure bet that happens to own the best launch business ever built."

"I think this is a $10 to $30 stock… I wouldn't get near this thing, I wouldn't even short it, because it's a meme stock." (Scott Galloway)

That AI framing got a surreal exclamation point this week. Per the Elon Musk Podcast, "SpaceX buys Cursor for 60 billion dollars" (Aug 15) (pundit) and the 20VC (Aug 20) (investor/pundit), SpaceX reportedly acquired the AI-coding startup Cursor for $60B, entirely in stock (~23x Cursor's ~$2.6B annualized revenue). The pundit show relayed an internal Musk claim that within ~5 years AI will make up 99% of SpaceX's value, the rockets reduced, in that telling, to "the trucks," and the satellite network to "the fiber-optic cable." Treat the Cursor deal as widely-reported-but-not-independently-confirmed here; the signal is that SpaceX is deliberately repositioning itself as an AI company, which is exactly what the bond market seems to be pricing.

4. The launch cadence keeps normalizing the absurd

Source: Astronomy Daily, "Tiny Robots, Alien Caves" (Aug 17), pundit · Elon Musk Podcast, "Texas Landowners Fight Starbase Annexation" (Aug 19), pundit

On Saturday, Aug 15, SpaceX launched two Falcon 9 rockets 38 minutes apart from opposite coasts, Cape Canaveral (carrying eight Globalstar satellites) at 9:12pm ET, then Vandenberg (a classified Space Force mission, USSF-366, likely on the Starshield national-security platform) at 9:50pm ET. That beat its own 65-minute turnaround record from 2024. The two boosters were on their 14th and 18th flights. SpaceX is now at 96 Falcon 9 launches in 2026 with four-plus months to go, roughly three-quarters of them building out Starlink.

Meanwhile the FAA has cleared Starbase for up to 25 Starship orbital launches, 25 Starship landings, and 25 booster landings per year, 75 major rocket events, or roughly one every 4–5 days. The relevance for everyone else in this letter: launch is becoming a cheap, abundant commodity, which is bullish for constellation builders (AST, Rocket Lab's customers) and brutal for anyone whose thesis rests on launch scarcity.

The debate: how big is direct-to-device, really?

This is the question the whole sector rides on, so let's steel-man both sides honestly.

The bull case (demand is real, inelastic, and the smart money is consolidating around it). Coverage gaps are a genuine, felt problem, the AST host opened his episode with a very relatable story of driving blind for 40 minutes in the Berkshires with one useless bar of signal. On the SpaceX side, the connectivity business is already throwing off serious cash, the Elon Musk Podcast (Aug 17) (pundit) cited $1.66B of quarterly operating profit from the connectivity segment alone, a "flawless" enterprise retention rate, and a $6B secure government-network backlog, with the eye-popping detail that some airline passengers now book less efficient connecting flights just to fly on Wi-Fi-equipped aircraft. That's pricing power. And the industry is voting with its wallet: over the past 18 months, essentially every stranded spectrum holder has been bought or partnered, Ligado, EchoStar's spectrum (to SpaceX), Globalstar (to Amazon), Iridium (to Rocket Lab), and now OmniSpace/Link (merged, with SES investing). When incumbents who spent years calling D2D a niche suddenly pay up for it, the TAM is being validated in real time.

The bear case (the numbers don't yet exist, and the valuations already assume they will). Nobody on-microphone this week could point to real direct-to-device revenue at scale, it's contracts, tests, and MOUs, not paying subscribers. T-Mobile's own CFO publicly argued satellite service is "limited" and "will never replace terrestrial." Galloway's harsher point is that the market has stopped pricing SpaceX on connectivity at all and is now pricing an AI-capex moonshot, SpaceX reportedly spent over $4 of AI capex for every $1 of new AI revenue last quarter, and its own projections (per the pundit show) imply a ~$105B cash-flow trough by decade's end requiring up to $350B of outside capital. For the pre-revenue pure-plays (AST especially), the bull case is the D2D TAM, which means if the timelines slip or ARPU disappoints, there's no legacy business to catch the fall. The honest synthesis: the demand is probably real; the timing and monetization are the entire ballgame, and today's prices give the operators very little room to be late.

Stocks in play

AST SpaceMobile (ASTS)

  • Bull: First documented T-Mobile linkage via the Grain test authority; a Grain operator decision due ~November with AST the obvious pick; a senior M&A dealmaker (Ozzy Ramos, ex-Lehman/Barclays/UBS) hired, hinting at more deals/JVs; a $1B+ Japanese sovereign subsidy behind the J-LEO joint venture. All per the AST SpaceMobile Podcast (investor).
  • Bear: Still pre-meaningful-revenue and cash-hungry; the "T-Mobile is coming" narrative is investor inference from FCC breadcrumbs, not an announcement; success is 100% levered to the D2D thesis holding.
  • Next catalyst: A formal T-Mobile agreement (the host guesses before year-end, after the three-carrier JV is inked); the next BlueBird launch; Grain's operator selection.

Rocket Lab (RKLB)

  • Bull: Record Q2, $2.36B backlog, ~$943M of 2026 defense wins, and an Iridium deal that buys cash flow + spectrum and pulls "space applications" revenue forward ~3 years; ~$48–49B market cap per Wall Street Wildlife.
  • Bear: Neutron still hasn't flown and keeps slipping; operating margin compressing under Neutron R&D; the Iridium deal flips the balance sheet toward net debt and needs Iridium-holder approval; a lot of manifest riding on an unproven rocket.
  • Next catalyst: A confirmed Neutron first-flight date; Iridium deal closing/approval; the promised 7investing full DCF revision.

SpaceX / Starlink (private)

  • Bull: Lockup passed as a non-event, ~40% rally, index-inclusion demand, relentless launch cadence (96 in 2026), a self-funding connectivity engine, and an AI-infrastructure option on top.
  • Bear: ~94x revenue at IPO, a "broken IPO" off its peak, a business now dependent on a colossal AI-capex bet funded by debt to 2056; Galloway's "$10–$30 stock" and "meme stock" call.
  • Next catalyst: Further staggered unlocks through December; whether the $100B ARR-by-year-end target (off a ~$30B June run rate) shows any traction; June 2027 "Elon day."

Iridium (IRDM) is in play only as Rocket Lab's takeout target: $871M FY2025 revenue, ~57% cash-flow margin, 66 satellites, being valued as the spectrum + cash flow that makes Rocket Lab's orbital-services pivot real. Next catalyst: shareholder vote / deal terms.

Read-throughs

  • Carrier partners (VZ, T, TMUS): No dedicated carrier episodes this week, but the read-through is the whole AST story. T-Mobile is the mover, the Grain test authority's T-Mobile clause plus Chris Sambar's hire suggest it's edging toward AST even as it keeps Starlink for now (Starlink runs on T-Mobile's 5×5 MHz mid-band and won't get EchoStar spectrum in hand until late 2027). AT&T and Verizon remain AST's anchor carriers, with exclusive deals said to survive the coming three-carrier joint venture.
  • Globalstar (GSAT): Quiet week on GSAT for commentary, no episodes on Apple's emergency SOS or the Amazon-ownership angle. But note the operational data point: eight Globalstar satellites rode SpaceX's Aug 15 Falcon 9, so the constellation replenishment is proceeding even as the investor narrative treats GSAT as an Amazon asset now.
  • EchoStar / Hughes (SATS): Thin, and not good. Per The Dan Rayburn Podcast (Aug 16) (pundit), EchoStar lost 80,000 Sling TV subs, 161,000 satellite-TV subs (Sling now just 1.7M), and 59,000 broadband subs, with total revenue down 4% YoY. The bull case for SATS was always the spectrum, which is heading to SpaceX; the operating business keeps eroding.
  • Launch & component suppliers: Quiet on ULA and Blue Origin/New Glenn this week (no dedicated coverage). Adjacent hits: Astronomy Daily (Aug 21) covered LandSpace landing China's first private reusable booster, a reminder the reusability moat is being chased globally; and The Pair Program (Aug 18) flagged a $50M Space Force contract for Northwood Space (ground comms) and Turion Space's $85M Series B in the space-domain-awareness niche.
  • Defense demand: Off-Nominal (Aug 20) treated Golden Dome as the sector's dominant procurement magnet (favoring Lockheed and L3Harris in its mock portfolio), and On Orbit (Aug 18) featured BlackSky CTO Patrick O'Neil (operator/insider) on how modern warfare is pulling sovereign customers toward commercial satellite intelligence, the same tailwind behind Rocket Lab's SBAMTI win.
  • SpaceX as sentiment anchor: The wealthy keep piling in, Bloomberg Intelligence (Aug 17) tallied $3.8B of family-office SpaceX bets, and Limitless (Aug 18) noted Gavin Baker's ~$4.7B position plus Nvidia's multi-billion-dollar stake. When SpaceX sentiment runs hot, ASTS and RKLB tend to trade with it, and vice versa on the way down.

What changed vs last week

  • AST: Last week was Q2 results and the J-LEO/Rakuten award. New this week: the FCC's Grain test authority and its T-Mobile compliance clause, a genuinely new, tradable data point on the missing carrier, plus the Sambar-to-T-Mobile move. The T-Mobile thread went from "speculation" to "documented breadcrumb."
  • Rocket Lab: Last week we had the $8B Iridium bid and Q2 preview. New this week: the actual Q2 print ($234M, backlog $2.36B), the $397M SBAMTI award confirming ~$943M of 2026 defense, and 7investing's reframing of Iridium as spectrum-plus-cash-flow pulling space-applications revenue to ~2028. Neutron timing: no change, still officially Q4 2026, still doubted.
  • SpaceX: Last week the story was the earnings shock and whether the Aug 6 lockup would matter. This week it's resolved: the lockup was a non-event and the stock rallied ~40%. The narrative has hardened around "leveraged AI-capex bet," punctuated by the reported $60B all-stock Cursor deal and the "AI = 99% of value" claim.
  • Globalstar: Still quiet on commentary (now several weeks running), but this week added an operational marker, 8 sats launched Aug 15.
  • EchoStar/Hughes: Broke its silence, but only via a bleak subscriber print. Iridium and the three-carrier JV definitive agreement remain podcast-quiet; ULA and Blue Origin stayed dark again.