SpaceX Raises $25 Billion in Debt as Tesla Locks Up $5 Billion European Storage Deal

A four-times-oversubscribed bond sale reframes SpaceX as investment-grade infrastructure while Tesla's 25 GWh NatPower agreement, a Finnish FSD review, and the Starfall demo push the empire deeper into energy, autonomy, and orbital services.

The most important thing that happened this week is not a rocket or a car—it is a credit rating. When Moody's, Fitch, and S&P all stamp SpaceX investment grade and the bond market answers a $25 billion offering with roughly $89 billion in orders, the signal is that capital markets have quietly reclassified a launch company as a durable infrastructure platform. That reclassification, more than any single product announcement, is what compounds. Meanwhile Tesla is doing the same trick on the physical side—turning Megapack backlog into multi-year European contracts—while the harder, unresolved questions of autonomy and orbital services move one incremental step closer to becoming real businesses.

Key Takeaways

• SpaceX priced $25 billion in five-tranche senior unsecured notes, drawing roughly $89 billion in demand—more than four times the offering.
• Coupons run from 5.35% to 6.65%, with maturities laddered from 2031 out to 2056, and proceeds mostly repay a bridge loan.
• Investment-grade ratings arrived from Moody's, Fitch, and S&P before pricing, alongside a disclosed $100.8 billion cash position as of June 19th.
• Tesla and NatPower signed a first phase worth up to $5 billion covering 25 GWh of Megapack across five projects in Italy and the UK.
• That single order equals more than half a year of Lathrop's stated ~40 GWh annual Megapack capacity, with the full program targeting 100+ GWh.
• Finland's Traficom is weighing early recognition of the Netherlands' FSD Supervised approval ahead of an October 2026 EU committee vote.
• Cybertruck became the only pickup to earn IIHS Top Safety Pick+ for 2026, avoiding every collision in pedestrian front crash tests.
• Musk confirmed "Starlink" as the name for a proposed constellation of one million AI satellites, though no specs or filings have followed.

---

Credit Markets Just Repriced What SpaceX Is

The number that matters is not the $25 billion raised but the $89 billion that showed up wanting in. A bond book that oversubscribes by more than four times means investors are competing to lend, and they are doing it at coupons between 5.35% and 6.65%—the kind of spread you extend to an infrastructure issuer, not a moonshot. The laddered maturities from 2031 to 2056 tell the same story: lenders are comfortable being repaid decades out.

This is what mispricing correction looks like when it happens to a private company. SpaceX operates in enormous incumbent markets—launch, connectivity, eventually compute—and has the one thing debt investors actually underwrite, which is a record of execution. The investment-grade ratings from all three major agencies are the formal version of a conclusion the order book had already reached.

Refinancing Is Not Growth Capital

The nuance that gets lost in a headline number is the use of proceeds. This raise primarily repays borrowings under a bridge loan facility, with the remainder going to fees and general corporate purposes. That is balance-sheet housekeeping, not a war chest for the next Starship variant. Treating it as fresh growth fuel would be a category error.

The more honest read is that SpaceX is terming out short-term debt into long-dated notes at attractive rates—a sign of financial maturity rather than a new spending program. With a disclosed $100.8 billion cash position, the company is not raising because it is starved; it is raising because the market is offering cheap, patient money and it would be foolish not to take it. What to watch next is whether the new notes trade well in the secondary market and whether a follow-on issuance appears with a clearer growth mandate attached.

Tesla Energy Is Still the Most Underpriced Line Item

The NatPower agreement is the clearest evidence yet that Megapack demand is structural, not promotional. A first phase of 25 GWh across five projects in Italy and the UK, worth up to $5 billion, commits more than half a year of Lathrop's stated ~40 GWh annual output to a single counterparty. The broader program targets over 100 GWh and potential revenue north of $15 billion across two decades.

The market keeps connecting AI electricity demand to grid storage too slowly, which is exactly why this segment stays cheap relative to its trajectory. Every data center buildout is a latent Megapack order, and deals like this one convert that latent demand into contracted backlog. The thesis is not complicated; it is just early.

The Real Test Is Margin, Not Backlog

Revenue growth alone was never the full energy thesis, and the skeptical read here is the correct one to hold in tension. A large Megapack contract can add only a modest slice to margin if it is priced as commodity hardware. Backlog is necessary but not sufficient.

The interesting wrinkle is Autobidder. NatPower will run its systems on Tesla's software platform, and Tesla is providing the full engineering, procurement, construction, hardware, software, and service stack. If that software-and-services layer meaningfully lifts the margin profile above bare metal shipments, the energy business stops being a hardware volume story and becomes something closer to recurring infrastructure revenue. The milestones that matter are binding delivery schedules, project-level approvals, and evidence that the software actually moves margins.

Regulation Has Become the Binding Constraint on Autonomy

Finland is a small market—roughly 6,500 eligible vehicles with current-generation hardware—which is precisely why the Traficom situation is more interesting than its size suggests. The agency is assessing whether to recognize the Netherlands' provisional type approval for FSD Supervised, and it may decide before the EU technical committee votes in October 2026. A single country moving early creates a live European reference case.

Traficom's language is notably positive, citing potential to reduce human-factor accidents and improve traffic flow, while being explicit that this is not automated driving and the driver must stay ready to intervene. The regulator is doing real diligence—winter driving, overtaking on curved roads, speed-offset behavior—which makes any eventual approval more durable as precedent. Investors fixated on robotaxi revenue timing are watching the wrong clock; every jurisdiction that lets FSD operate makes each Tesla more differentiated long before any robotaxi network exists.

Safety Ratings and Insurance Are Quietly Compounding

The Cybertruck earning IIHS Top Safety Pick+ for 2026—the only pickup to do so, avoiding every collision in day and night pedestrian tests—is not a vanity award. Independent safety validation feeds directly into demand, insurance economics, and the broader narrative that Tesla's hardware is measurably safer. Those are the inputs that move total cost of ownership.

The insurance angle sharpens this. Lemonade expanded its autonomous-car product to Colorado, offering Tesla drivers 50% off every mile driven on FSD Supervised, using permissioned vehicle data to separate autonomous miles from manual ones. When an insurer is willing to price supervised-autonomy miles as materially safer, it is putting money behind a claim Tesla has been making for years—and it turns safety into a recurring cost advantage rather than a marketing line.

Starlink and Starfall: Optionality, Not Yet Business

Two SpaceX stories this week are best understood as options rather than operations. Musk confirmed "Starlink" as the name for a proposed constellation of one million AI satellites, and X AI filed a trademark spanning software, communications, education, and AI services. The strategic logic is clean—SpaceX brings launch cadence and Starlink distribution, X AI brings model development and compute demand—but a name is not an architecture. There are no specs, launch schedules, regulatory milestones, or customers, and a constellation at that scale runs straight into questions of power, thermal management, orbital debris, and spectrum.

Starfall is further along but still a demo. SpaceX launched the vehicle on Falcon 9 and confirmed orbital deployment, describing it as affordable, routine access to microgravity for research and in-space manufacturing, with a planned Pacific splashdown. The prize is moving SpaceX from launch provider to repeatable orbital-return services operator—a specialized market Dragon does not target. But controlled re-entry, ocean recovery, and repeat flights must work reliably and cheaply before anyone builds a workflow around it. The immediate thing to watch is splashdown performance, then pricing and cadence.

The Through-Line: Turning Ambition Into Underwritable Assets

Step back and the week has a single theme. SpaceX converted execution history into an investment-grade balance sheet. Tesla converted manufacturing capacity into contracted European revenue. Finland is inching toward converting a Dutch approval into European regulatory precedent. In each case the pattern is the same—taking something speculative and making it legible enough for a lender, a regulator, or a utility to underwrite.

The Starlink constellation and Starfall sit at the opposite end of that spectrum, still pure optionality with no filings or customers to anchor them. That contrast is the useful frame for the whole enterprise right now: a maturing core throwing off cheap capital and contracted cash flow, funding a frontier that remains undefined. The discipline for anyone watching is to price the two differently—and not to let the excitement of the frontier obscure how much the core has quietly de-risked.