Newsletter · · Ashutosh Agarwal

SpaceX AI Spending Spooks the Market as Rocket Lab Bids for Iridium - The Satellite & Space-Comms Race - Week of August 14, 2026

A synthesis of what space and satellite investors and analysts said on podcasts for the week of August 14, 2026, as SpaceX's debut earnings were swamped by a $15.8 billion AI-infrastructure bill, AST SpaceMobile executed a strong quarter with up to $1 billion of Japanese government funding, and Rocket Lab bid $8 billion for Iridium.

The Satellite & Space-Comms Race

Week of August 14, 2026: SpaceX AI Spending Spooks the Market as Rocket Lab Bids for Iridium

Two weeks ago this whole complex was getting dragged around by a blown-up hedge fund. This week the story flipped back to fundamentals, and it turns out the fundamentals had a lot to say. SpaceX filed its first-ever earnings report as a public company and the market recoiled at the size of its AI bill. AST SpaceMobile launched three more satellites, held a genuinely strong quarter, and locked in up to a billion dollars of government money in Japan. And Rocket Lab stopped being a rounding error in the news and did something bold: it agreed to buy Iridium for $8 billion. Grab a coffee. This one earned its length.

(Speaker tags below: OPERATOR/INSIDER = company management; INVESTOR = someone with money on the line; PUNDIT = commentator or independent analyst.)

TL;DR

  • SpaceX's debut earnings were a "believers, not investors" moment. Revenue jumped 92% to $7.81 billion (USD) and the loss came in far smaller than feared, but a single-quarter $15.8 billion AI-infrastructure spend swamped everything, and the stock traded double digits lower before clawing back toward its ~$135 IPO price by week's end.
  • AST SpaceMobile quietly executed. Q2 revenue of $31.5 million (USD), full-year guidance of $150–200 million reaffirmed, up to ~$1 billion (USD) of non-dilutive Japanese government money via Rakuten, three new US government contracts worth $100 million+, and three fresh BlueBird satellites launched August 5.
  • Rocket Lab is buying Iridium for $8 billion to bolt a real, cash-generating services layer onto its rockets, and, per the earnings call, spent far more airtime talking about the still-grounded Neutron rocket ("Neutron" said 61 times) than about the rocket actually making money today.

What's new

1. SpaceX's first earnings report: a revenue beat drowned out by a $15.8 billion AI bill

This was the dominant story of the week by podcast volume, more than two dozen shows dissected it. The setup: SpaceX went public in June and the stock had spent the summer below its offer price, so a good print was supposed to be the catalyst that got it moving again. It wasn't.

The numbers, as laid out on the Elon Musk Podcast (Aug 7, PUNDIT): revenue of $7.81 billion (USD), up 92%, with the net loss narrowing to just $541 million, a fraction of the ~$2.12 billion loss the market had penciled in. Normally that's a "turning the corner" quarter. Instead the stock sank about 11% pre-market, because sitting on the report was a $15.8 billion capital expenditure on AI infrastructure, spent in a single quarter, double the prior quarter's AI spend, with management guiding for that pace to continue for at least two more quarters.

"The company is spending $6.18 on AI infrastructure for every $1 of AI revenue generated. That ratio is exactly why the stock is dropping."

Elon Musk Podcast, Aug 7

The hosts explained the bull rebuttal fairly: SpaceX isn't building chatbots, it's becoming "the landlord for the AI industry," renting out NVIDIA-powered compute, and AI revenue already tripled quarter-over-quarter to $2.56 billion (USD). CFO Brett Johnson defended the spend on the call by pointing to a sub-one-year payback period on AI compute deployments, reframing a scary $15.8 billion outlay as a fast-returning asset rather than "a hole in the ground." The plan is to grow power capacity from ~2 gigawatts toward ~10 gigawatts, the output of several nuclear reactors, dedicated entirely to computing.

The bears got the sharper lines. On The Real Eisman Playbook (Aug 7, INVESTOR), the host flagged that total capex jumped to $18.4 billion from $10.1 billion in the March quarter, and that Starlink, "the only profitable business," added subscribers to 12 million, just shy of the 12.19 million expected. His bottom line:

"Elon Musk has a moat built by his personality. He has believers, not investors."

He pegged the valuation at roughly 54x annualized revenue ($7.8 billion × 4 = $31.2 billion, against a ~$1.6 trillion market cap), "a tad expensive."

Why it moves numbers: SpaceX is the sentiment anchor for the entire sector. When the biggest, most-admired name gets punished for spending ahead of revenue, every pre-revenue satellite story (hello, ASTS) gets re-underwritten with a more skeptical pen. The flip side, per CNBC's Fast Money (Aug 7, PUNDIT): by Friday, Argus had upgraded SpaceX to buy with a $160 price target and the stock had rebounded ~16% back toward its IPO price. The panic, it seems, had a short shelf life.

2. AST SpaceMobile's Q2: revenue small, everything else large

AST held its Q2 2026 earnings call on August 10 (OPERATOR/INSIDER, Chairman/CEO Abel Avalon, President Scott Wisniewski, CFO Andy Johnson). The revenue line is still tiny for a company this size, $31.5 million (USD), but management reaffirmed full-year 2026 guidance of $150–200 million and stacked up a lot of substance behind it:

  • Japan, funded: an award (pending approvals) with long-time partner Rakuten to join Japan's "GLEO" low-Earth-orbit infrastructure project, worth up to ~$1 billion (USD) in non-dilutive, non-debt government capital. That is free money to build, aimed at the Japanese and Asian markets.
  • Government traction: three new US government contract awards with $100 million+ in funded near-term value across 2026–2027, plus continued FirstNet emergency-network work. Management framed this as small development contracts graduating into larger ones: "ahead of still larger operationalization… through programs of record."
  • Backlog and balance sheet: revenue backlog up to ~$1.3 billion (USD) in aggregated contracted revenue, and a balance sheet of more than $3.7 billion (USD), further bolstered by July's convertible-debt offering.
  • Spectrum: on track to roughly 100 MHz of US spectrum (a mix of partner-provided and AST's own), with satellites able to tune across ~1,150 MHz, plus 3,900+ patents and pending claims.
  • Deployment: three new BlueBirds (14–16) are "ready to ship shortly"; the company is now building through BlueBird 46 and targeting ~45 satellites in orbit by early 2027, on the way to a 100+ satellite constellation at a cost of roughly $21–23 million (USD) per satellite. It reiterated the goal of six fully assembled satellites per month and unveiled another 400,000 sq ft of manufacturing in Midland, Texas.
  • The cost of all this: Q2 capex was ~$610 million (USD) (up from ~$257 million in Q1) and adjusted operating expenses were $119.1 million (USD). This is a company spending hard.

The most useful outside read came from Anpanman (Aug 11, INVESTOR), who argued the quarter was "very bullish" even though the stock traded in sympathy with the broader space sell-off. His standout data point wasn't a financial metric, it was a hire:

"You don't have a vice chair of investment banking join a shit co."

AST brought on Ozzy Ramos, most recently vice chair of UBS's investment banking division and previously head of Barclays' communications, cable and satellite coverage. Anpanman's read: you don't recruit a banker of that seniority unless there's serious strategic and M&A/joint-venture work coming: the Rakuten JV, a US JV, European deals, and possibly spectrum tie-ups.

Why it moves numbers: AST is the purest direct-to-device (D2D) bet, putting a normal, unmodified phone onto a satellite. The Rakuten billion de-risks the funding overhang that has haunted every pre-revenue space name, and the government contracts hint at a second, higher-margin revenue engine (defense, radar, edge compute) that isn't in most models yet.

3. Rocket Lab bids $8 billion for Iridium, and finally shows up in the conversation

After two straight weeks of near-silence, Rocket Lab came back with a bang. On Chip Stock Investor (Aug 7, INVESTOR), Nicholas and Kasey Rossolillo broke down RKLB's proposed $8 billion (USD) acquisition of Iridium Communications, a mix of new stock and cash that flips Rocket Lab from net cash to net debt.

The logic is vertical integration: Rocket Lab already builds satellite components, builds satellites, and launches them (the small Electron rocket today, the bigger Neutron by end-2026, in theory). Iridium bolts on the missing top layer: an operating 66-satellite constellation with reliable, global L-band voice/text/data used by maritime, aviation, and government customers, plus, crucially, cash flow. The pro-forma math the hosts walked through:

  • Combined trailing-12-month revenue of ~$1.6 billion (USD), with $884 million from Iridium (the bigger, mature business).
  • Iridium generated +$288 million of free cash flow and +$423 million of EBITDA; Rocket Lab burned -$316 million of free cash flow and posted -$173 million EBITDA. Combined, the group is close to break-even: "Iridium essentially buys them some free cash flow."
  • Even so, at a combined ~$45 billion market cap / ~$50 billion enterprise value, this trades at roughly 200x EV/EBITDA. To justify that, the hosts calculated Rocket Lab would need ~35% annual EBITDA growth for a decade, taking EBITDA from ~$250 million to ~$5 billion by the mid-2030s. For context, SpaceX's 2025 EBITDA was just over $4 billion.

The earnings-call color came from Motley Fool Hidden Gems Investing (Aug 12, INVESTOR). Rocket Lab's Q2 revenue beat (a record, up ~62%) and backlog more than doubled year-over-year, but the bottom-line loss was wider than expected, and the priorities were telling:

"The word neutron appeared in the conference call 61 times… and only 17 times for electron, you know, the rocket they're actually using to generate those earnings."

Host Travis Hoium noted the stock is up ~1,100% over three years and carries a ~$50 billion market cap against just $2.36 billion of backlog, so any hint of Neutron slipping gets punished.

Why it moves numbers: This reframes Rocket Lab from "launch company hoping to grow" into "vertically integrated space services company with real cash flow." It also puts Iridium in play as an acquired asset, and independent expert Tim Farrar of TMF Associates, speaking on Prof G Markets (Aug 12, PUNDIT), noted every one of these names (SpaceX, Rocket Lab, AST) is now chasing the same three-layer model: rockets, satellites, and the service on top.

4. AST's write-off and the launch-access scramble get named out loud

The clearest independent framing of AST's quarter came from Tim Farrar on Prof G Markets (Aug 12, PUNDIT). The hosts noted AST's losses more than doubled to $231 million (USD) in the quarter, "a big chunk of that" being a $126 million (USD) write-off from a satellite that Blue Origin had launched into the wrong orbit earlier this year. Farrar's blunt take on why AST raised over a billion dollars and is eyeing the launch business itself:

"They just raised over a billion dollars… to try and move into the launch business because they're struggling to get launches from third parties… and now they're scrambling for launches."

He tied it together: AST's launch woes were "amplified when the New Glenn rocket exploded on the pad" (more on that below), leaving the company hunting for rides to orbit. It's the counterpoint to the bull case: execution and launch access remain the real risk, not the demand story.

5. Blue Origin names the culprit for its New Glenn explosion

On This Week in Space (Aug 7, PUNDIT, Rod Pyle and Space.com's Tariq Malik), the hosts reported that Blue Origin has pinned the cause of its New Glenn destruction (a first-stage test that blew up on the pad back in late May, taking out pad structure) on a faulty oxygen valve in one of the BE-4 engines. CEO Dave Limp has been unusually transparent, posting investigation and rebuild updates, "very refreshing," the hosts noted, for a company not historically forthcoming.

Why it moves numbers: New Glenn is a needed launch option for the whole industry (and for NASA's moon plans). Its grounding is precisely why AST took a satellite write-off and is spending to secure its own path to orbit, and why launch capacity, not satellite production, has become this year's swing risk for AST.

The debate: how big is the direct-to-device (D2D) market, really?

This is the question the whole universe hinges on: will normal people actually pay for their phone to talk to a satellite?

The bull steel-man. D2D is a brand-new connectivity layer on top of the ~6 billion phones already on Earth, and AST is the partner of choice: 60+ mobile-network operators covering 3+ billion subscribers, from AT&T and Verizon to Vodafone, Rakuten, Saudi Telecom (STC), Bell Canada and Deutsche Telekom (AST SpaceMobile Q2 call, Aug 10, OPERATOR). On The Kook Report (Aug 8, INVESTOR), the case widens beyond consumer phones: spectral efficiency improving from 2 to 10 bits per hertz, a Saudi Telecom deal with $1 billion (USD) of minimum contracted revenue, and a growing menu of "non-communications" uses (radar, emergency response, IoT, defense) that could dwarf the phone-connectivity market. The government is leaning in: on Kook's Weekly (Aug 8, INVESTOR), Kook flagged FCC Chairman Brendan Carr publicly "clearing the path" for D2D and freeing up spectrum, and AST's own Jennifer Manor being named to the FCC's Technological Advisory Council, a literal seat at the rule-writing table.

The bear steel-man. Nobody has proven anyone will pay for this. On Motley Fool Hidden Gems (Aug 12, INVESTOR), Travis Hoium put it plainly:

"Are people actually going to sign up for a satellite connection for their phone? Is that going to be included in your plan if you're on Verizon or AT&T? What's the real business model behind it?… We're getting really close where these companies are going to have to start showing revenue, margin, whether they have pricing power or not."

And it's not a one-horse race. Chip Stock Investor noted Iridium can already offer a form of D2D this year with software updates to its existing constellation, and SpaceX's Starlink Direct-to-Cell is scaling fast. The bear case isn't that the market is small, it's that it's unproven, competitive, and being funded today with real cash burn against revenue that's still mostly a promise.

Where I land: the government and defense demand is the underappreciated swing factor. If D2D were only ever a consumer add-on, the bears would probably win on economics. The moment it becomes national-security infrastructure (resilient PNT, secure comms, radar) the willingness-to-pay problem largely goes away, because the customer is the Pentagon, not a price-sensitive consumer. That's the part of AST's story that isn't in most spreadsheets yet.

Stocks in play

AST SpaceMobile (ASTS) (OPERATOR/INSIDER + INVESTOR coverage, heavy week)

  • Bull: Reaffirmed $150–200M FY26 revenue; ~$1B non-dilutive Rakuten/Japan money; $100M+ in new US government contracts; ~$1.3B backlog; $3.7B+ balance sheet; three satellites launched Aug 5, targeting ~45 in orbit by early 2027; a UBS vice-chair-caliber banker just joined.
  • Bear: Still only $31.5M of quarterly revenue against ~$610M of quarterly capex; a $126M write-off on a lost satellite; genuine launch-access risk after Blue Origin's failure; the D2D consumer business model is unproven.
  • Next catalyst: The next BlueBird launch (Kook's estimate is early-to-late September); formal terms on the Rakuten JV and the US carrier JV; and details on those three government contracts.

Rocket Lab (RKLB) (INVESTOR coverage, back in the spotlight)

  • Bull: $8B Iridium deal adds a cash-generative services layer and flips the group toward break-even; record Q2 revenue (~+62%); backlog more than doubled; up ~1,100% over three years.
  • Bear: Flips from net cash to net debt; ~200x EV/EBITDA needs ~35% EBITDA growth for a decade to justify; Neutron still hasn't flown and the call obsessed over it; wider-than-expected loss.
  • Next catalyst: Neutron's first flight (management still says end-2026; multiple podcasters expect 2027) and Iridium deal closing/regulatory progress (needs ~35% Iridium shareholder approval, per Chip Stock Investor).

SpaceX (private, publicly traded shares) (PUNDIT + INVESTOR coverage, dominant volume)

  • Bull: 92% revenue growth; AI revenue tripled to $2.56B; CFO claims sub-one-year payback on compute; Argus upgraded to buy with a $160 target; Starlink profitable at 12M subs.
  • Bear: $15.8B single-quarter AI capex with a 6-to-1 spend-to-revenue ratio; ~54x revenue; "believers, not investors"; the payoff (data centers in space) is years away: WSJ cited a Moffett-Nathanson estimate of 143 Starship launches to put a single gigawatt of computing into orbit.
  • Next catalyst: Further share-lockup unlocks through the fall; whether AI revenue "exceeds other SpaceX revenue in September," as management floated on Squawk on the Street (Aug 12, PUNDIT).

Iridium (IRDM) (covered only as an acquisition target this week)

  • Bull/Bear: The story is now the Rocket Lab takeout: a mature, cash-generative, government-heavy L-band constellation ($884M revenue, +$288M FCF) being folded into a high-growth acquirer. Standalone thesis has effectively been replaced by deal terms.
  • Next catalyst: Deal approval and disclosed exchange terms.

Read-throughs

  • Carrier partners (VZ, T, TMUS): No fresh dedicated carrier commentary this week beyond AST's own disclosures, but AST management reaffirmed AT&T, Verizon and Deutsche Telekom relationships, and Anpanman raised the intriguing idea that the emerging carrier "joint venture" may be a distinct fourth entity that AST works with alongside the big three, potentially holding spectrum (like AST's Ligado/"Legato" holdings) and expanding into IoT. Quiet week on a three-carrier JV definitive agreement specifically, no podcast surfaced one.
  • Globalstar (GSAT) / Apple emergency SOS: Quiet week on Globalstar, no GSAT-specific podcast discussion surfaced, now roughly the fifth straight near-silent week. The relevant context came second-hand: on Anpanman's episode and Chip Stock Investor, hosts noted Amazon won Globalstar (after both SpaceX and Rocket Lab reportedly bid but couldn't get Apple's cooperation), which is exactly why Rocket Lab "moved on" to Iridium. Globalstar is increasingly discussed as an Amazon asset rather than a standalone story.
  • EchoStar / Hughes (SATS): Quiet week, no dedicated coverage.
  • Launch & component suppliers: Blue Origin dominated (New Glenn root cause, above). On the SpaceX side, Moonshots with Peter Diamandis (Aug 8, PUNDIT) detailed the Aug 4 announcement that SpaceX and NVIDIA are jointly designing the "StarMine" orbital-data-center payload (each satellite carrying NVIDIA Rubin GPUs and Vera CPUs, with first deployment pulled forward to 2027), and a Reuters report that SpaceX and Tesla will invest $16.8 billion (USD) in a "TerraFab" semiconductor complex. Redwire and Intuitive Machines had no earnings-relevant coverage this week.
  • Defense / national security (SDA, NRO, Golden Dome): Quiet week on dedicated defense-demand episodes, though Motley Fool flagged the Pentagon's Golden Dome as a major reason investors are excited about space broadly, and AST's government contracts and PNT ambitions keep the thread alive inside its own coverage.
  • SpaceX private-market valuation as sentiment anchor: Still the tell for the whole group. On Money Meets Medicine (Aug 12, PUNDIT), advisor Justin recapped the June 11 IPO at $135/share raising ~$75 billion (USD), the largest ever, 30% to retail, with 2025 net losses of $4.9 billion and ~61% of 2025 revenue from Starlink, and criticized index providers for fast-tracking SpaceX's inclusion. The read-through: as SpaceX goes, so goes sentiment for ASTS and RKLB.

What changed vs last week

Last week was all anticipation: a preview of SpaceX's Aug 4 earnings and Aug 6 lockup, AST's Aug 5 launch and Aug 10 earnings. This week, every one of those events actually happened:

  • SpaceX earnings landed, mixed-to-disappointing on the AI spend, but the stock recovered toward its IPO price by Friday (Argus $160 target), and the 911 million-share lockup unlock on Aug 6 proved to be a non-event, "a red herring," per WSJ's hosts, because it was well telegraphed. That contradicts last week's fear of a lockup-driven leg down.
  • AST executed: the Aug 5 launch put three BlueBirds up successfully; the Aug 10 call reaffirmed guidance and confirmed the ~$1B Rakuten money; but it also revealed the $126M satellite write-off and made the launch-access problem explicit.
  • Rocket Lab broke its two-week silence in style with the $8B Iridium bid and an earnings report, resolving last week's "retrieval gap vs. true lull" question decisively in favor of "true lull, now over."
  • Blue Origin's New Glenn root cause is now known (oxygen valve / BE-4 engine), versus last week's open "awaiting root cause."
  • Still quiet: Globalstar, EchoStar/Hughes, and a formal three-carrier JV agreement, same as last week. The Situational Awareness hedge-fund blowup that drove everything two weeks ago has faded from the conversation as fundamentals reasserted themselves.