SpaceX +8%, Nvidia's $20B deal sued, OpenAI hits pause

Good morning, crowd. ☕

Quick game before we start. 🎯

Morgan Stanley's top SpaceX analyst asked a room of 40 clients a simple question: "Who here owns SpaceX stock?" How many hands went up?

• 🅰️ 31

• 🅱️ 12

• 🅲️ Zero

Hold that thought. The answer is in story #1.

📋 Today's 60-second summary

• 🚀 SpaceX jumped almost 8% and Elon Musk is a trillionaire again

• ⚖️ Ex-Groq engineers sued over Nvidia's $20B deal, saying common stockholders got squeezed

• 🧒 A 19-year-old raised $11M from a16z for a $3,499 AI computer with no screen

• 🛑 OpenAI hit the brakes: one model scrapped, training paused, IPO pushed to 2027

⏱️ Read time: 4 minutes

The ice cream shop that makes money when it's cold

28 Wishes sells ice cream in Los Angeles. Below 70°F, sales fall about 20%. The weather is out of their hands. Rent isn't.

So the owners started putting about $20 a day into weather markets, taking the cold side. The days that keep customers away now pay something back.

This is hedging. Big companies have done it for decades, buying protection against bad weather, fuel spikes and rising rates. It used to take a broker, a trading desk, and an order size no corner shop could meet.

opens it up. Contracts on weather, fuel prices, inflation, tariffs and regulation, starting at a few dollars. Take a position on the outcome that would hurt you. If it hits, the payout softens it. If it doesn't, the contract expires and the good month was the point.

NEWS OF THE WEEK

Space X

⚡ TL;DR

• SpaceX stock rose almost 8% on Monday to $171.09, its highest close since mid-June

• Morgan Stanley called the stock cheap and kept a $300 price target

• After a brutal post-IPO slump, the stock is up 58% from its August low

SpaceX shares rose almost 8% on Monday, closing at $171.09, the highest level since the weeks right after its record IPO in June. The rally pushed Elon Musk's net worth back above $1 trillion.

The trigger was a Sunday note from Morgan Stanley analyst Adam Jonas, who called SpaceX cheap and kept a $300 price target, about 75% above Monday's close.

🧠 Quiz answer: It was 🅲️, zero. Jonas said that when he asked 40 clients who held the stock, not a single hand went up. That's his whole bull case in one sentence: the big money is under-owned.

🔢 The numbers:

• $171.09: Monday's close

• $300: Morgan Stanley's price target

• +58%: rebound from the August low

• ~50%: how far shares fell from the post-IPO peak by late July

• $7.81B: Q2 revenue, which beat estimates, but with a 9 cent loss per share

• ~30x projected 2028 operating earnings, versus about 16x for mega-cap AI peers

Here's the thing. Jonas admits SpaceX looks expensive on simple measures. His argument is that once you adjust for growth, it trades about 40% below its AI peers. He's betting on three catalysts: new AI products, Starship progress and more Neocloud contracts, where SpaceX rents computing power to other companies. He specifically suggested buying before the next Starship test flight and before third-quarter earnings, expected in late October.

But here's the risk. Morgan Stanley also models about $80 billion a year in new net debt to fund the buildout. And a failed Starship test could hit the stock fast.

If you read our Starship story on October 1, you know why the rocket matters so much: Starship is what makes Starlink V3, and SpaceX's space data center plans, affordable.

🎯 What to do with this:

• Pre-IPO investors: SpaceX is the clearest case study of the year. A huge day-one pop, a near-50% drawdown, then a 58% rebound. Private market "wins" can whipsaw for months after listing. Plan your holding period before you buy.

• Founders: the market rewards a story it can model. SpaceX's rebound came from a clear list of catalysts with dates attached.

🗳️ POLL: Would you buy SpaceX at $171?

• 🚀 Yes, $300 is coming

• ⏳ Wait for the next Starship test

• 🫧 No, it's still too expensive

Groq

⚡ TL;DR

• Two former Groq engineers sued Groq's board, saying the deal squeezed out stockholders

• The $20B was split into a $17B license and $3B in Nvidia stock for engineers who moved over

• If this structure holds up, it changes what an "exit" means for startup employees

On paper, Groq had one of the biggest exits in AI chip history: a roughly $20 billion deal with Nvidia. Now two of its former engineers say most stockholders were left behind.

Joshua Rubin and Benjamin Serebrin filed a proposed class action in Delaware's Court of Chancery on October 2. They allege that Groq's board approved the deal without a required stockholder vote, gave stockholders a lowball price, and that a majority of the board had conflicts of interest.

🔢 How the $20 billion was split (per the lawsuit):

• $17B: a payment for a license Nvidia described as non-exclusive

• $3B: a pool of Nvidia stock (RSUs) for selected Groq engineers who joined Nvidia

• 150 to 200: Groq engineers who moved to Nvidia

• ~$3.5B: what the remaining Groq was later valued at in another funding round

Here's what's wild. Because the deal was structured as a license plus hiring, rather than an acquisition of the company, the plaintiffs argue stockholders never got a vote. The lawsuit also says the $17 billion was treated as taxable licensing income for Groq, another hit to what's left for shareholders.

Groq called the lawsuit meritless and said it will defend itself vigorously. The deal is also reportedly being looked at by the Justice Department.

Big Tech has used this "license and hire" structure several times to get a startup's technology and team without a formal acquisition. This is the first major legal test of whether it shortchanges the people who aren't in the room.

🕵️ SPOT THE RED FLAG: Which of these should worry a startup employee most?

• 🅰️ The deal was called a "license," not a sale

• 🅱️ Only selected engineers got the stock pool

• 🅲️ There was no stockholder vote

• 🅳️ All of the above

(Spoiler: it's 🅳️. Together, they decide whether your shares are worth anything when your company "exits.")

🎯 What to do with this:

• Employees with startup equity: ask what happens to your shares in a licensing deal, not just an acquisition. Read the "change of control" definition in your stock plan.

• Founders: if you ever structure a license-and-hire deal, involve your stockholders early. Lawsuits like this one can shadow a company, and a founder's reputation, for years.

• Private market investors: your protection is in the documents. Check what counts as a sale and when you get a vote.

Ghost

⚡ TL;DR

• Ghost, led by 19-year-old Zain Javaid, raised an $11M seed led by Andreessen Horowitz

• Its first product, Core, is a $3,499 screenless computer that runs personal AI agents

• The pitch: your AI runs locally, with no subscription

Zain Javaid wanted to be a quant trader. Then personal AI caught his attention. On Monday, his startup Ghost came out of stealth after a year of building, with an $11 million seed round led by a16z.

The founding team is young, even by Silicon Valley standards: Javaid and Nicholas Chua are both 19, Yifei Chen is 24 and Gautam Sharda is 23.

🖥️ What Core is:

• A $3,499 box with no screen, controlled from a phone app or by voice

• An Nvidia RTX Pro 4000 Blackwell GPU inside, built in San Francisco

• AI models and your personal memory run locally, and data is encrypted so even Ghost can't read it

• No subscription: one price, you own it

• Shipping is expected to start October 31

So why not just buy a Mac Mini? Javaid's answer: the Mac Mini is a general-purpose computer you have to configure. Core is built only for running AI agents, with the models already installed.

Here's why this matters beyond the headline. Every big AI company wants you on a monthly plan. OpenAI just launched a $500/month tier. Ghost is betting a growing group of people would rather pay once and keep their data at home.

🎯 What to do with this:

• Founders: watch the "private AI" trend. Privacy and ownership are becoming selling points people will pay a premium for.

• Investors: $11M is a modest seed for AI hardware. The real test is whether Core ships on time and people keep using it after the first month.

🗳️ POLL: Would you pay $3,499 for a private AI computer?

• ✅ Yes, I want my data at home

• 💸 Maybe, at half the price

• ☁️ No, the cloud is fine

⚡ TL;DR

• OpenAI scrapped its GPT-6.1 Astra model after it failed internal alignment tests

• It also paused training of its latest models, its second pause in three months

• Sam Altman says the IPO won't happen until 2027

The company that set the pace of the AI race is now the one tapping the brakes.

🗓️ What happened, in order:

• Sept 12: Sam Altman told Fortune that OpenAI won't go public in 2026 and that an IPO would be ill-advised this year, citing AI safety

• Sept 28: OpenAI shelved GPT-6.1 Astra after internal tests showed it acting outside its authorized scope

• Sept 29: at its DevDay event, OpenAI launched GPT-6.1 Sol at $2 / $10 per million tokens instead, plus its always-on Dots agents and a $500/month plan

• Oct 1: OpenAI reportedly parted ways with three safety-team researchers, saying they mishandled sensitive information outside company procedures

• That week: OpenAI paused training of its most capable models to investigate agents acting beyond their assigned tasks, its second pause in three months

According to the Wall Street Journal, Astra performed poorly on tests measuring alignment, meaning whether the model does what it's supposed to do and nothing else.

But the business isn't standing still. A day after DevDay, OpenAI said GPT-6.1 Sol was seeing the highest demand it had ever recorded. The company also plans to start testing visual ads in ChatGPT in the US.

Here's the private markets angle. OpenAI is reportedly raising a $30 billion bridge at a $1.4 trillion valuation, as we covered in Saturday's edition. Safety is now a line item in that valuation. Every pause delays revenue and pushes out the IPO, but skipping one could cost far more if something goes wrong.

Meanwhile, Anthropic appears to be moving ahead with its own listing. If it goes first, it sets the price benchmark for every frontier AI lab, OpenAI included. We broke down its filing on Sunday.

🎯 What to do with this:

• Investors in AI secondaries: "IPO in 2027" means a longer hold. Price that time into anything you buy now.

• Founders building on OpenAI: expect more pauses and model changes. Avoid depending on a single model provider.

🗳️ POLL: Is OpenAI's slowdown smart or scary?

• 🧠 Smart, safety first

• 😬 Scary, they're losing the race

• 💰 Neither, it's about getting the IPO price right

⚡ Quick Hits

• 🟢 Nvidia hit record highs after Morgan Stanley reinstated it as its top semiconductor pick

• 🧠 Cerebras stock popped 9% after Sam Altman called the chipmaker a close partner

• 📈 The Nasdaq and the S&P 500 tech sector closed at records on Monday

• 🎤 TechCrunch Disrupt runs October 13 to 15 in San Francisco, where the Startup Battlefield winner takes home $100,000 in equity-free funding

💬 What This Week Tells Us

Every story today is about the gap between the headline number and what people actually get.

• 🚀 SpaceX's IPO pop looked like a win, until the stock fell by about half

• ⚖️ Groq's $20B exit looked like a jackpot, until you check who got paid

• 🧒 Ghost's $3,499 price looks steep, until you compare it with years of subscriptions

• 🛑 OpenAI's $1.4T valuation looks locked in, until safety pushes the IPO back

📣 Your turn

If you had $10,000 today, where would it go?

• SpaceX stock at $171

• A seed check into the next Ghost

• Cash, waiting for the Anthropic IPO

Hit reply with your pick and one sentence on why. We'll share the best answers in Sunday's edition. 👀

📨 Know someone with startup stock options? Forward them the Groq story. It could save them a nasty surprise.

Track the biggest private companies on our Private Company Index.

Hunt accordingly. ☕

Jose, Alberto & the Founderscrowd Team