|
LIVE FROM THE EMPIRE · JULY 29, 2026
Grok is shipping faster than the news cycle.
I opened X this morning and saw the next two model numbers before most people had finished testing the last one. Grok 4.6 and 4.7 are already being teased. Cybercab production has started. Tesla and SpaceX keep finding new reasons to work as one system. The signal today is not one product. It is speed.
|
|
GROK
4.6 + 4.7 teased
The roadmap is moving before 4.5 feels old.
|
|
|
CYBERCAB
Production started
Tesla says the dedicated robotaxi is now being built in Texas.
|
|
|
THE OVERLAP
Tesla × SpaceX
Batteries, chips, AI, manufacturing, and networks are converging.
|
|
MUSK WEALTH TRACKER
|
LATEST VERIFIED FORBES FIGURE · JULY 27
$715.6B
|
−$9.5B vs. prior day
|
Forbes still showed Musk at number one in its latest indexed update. The live figure had not refreshed for July 28 when this edition was prepared, so I am keeping the last verified number instead of inventing a cleaner one. Tesla and SpaceX shares both moved again on Tuesday. That means the true live total may already be different.
Source: Forbes Real-Time Billionaires. Private holdings and stock-linked estimates can change quickly.
|
ACROSS THE EMPIRE
|
Elon is already talking about Grok 4.6 and 4.7
The newest signal on X is simple. More model releases are coming. Grok 4.5 only arrived in mid-July, built for code, agent tasks, and knowledge work. The next labels are already on the board.
A model number is not proof of progress. Benchmarks, cost, speed, and real user work still matter. But the cadence matters too. xAI is trying to make a new release feel normal, not rare.
Why this matters: Fast iteration compounds. Each release brings more use. More use exposes more failures. Those failures become the next training set. The winner may not be the lab with one perfect launch. It may be the lab that learns fastest between launches.
|
|
Cybercab has crossed from slide deck to factory line
Tesla says production of the dedicated Cybercab has begun in Texas. This is the two-seat vehicle built around autonomous service, not a normal car with a robotaxi option added later.
The design removes the steering wheel and pedals from the core idea. That makes the vehicle cheaper and simpler only if the driving system is ready. The software has to carry the whole product.
Why this matters: Starting production changes the debate. The question is no longer whether Tesla can draw a robotaxi. It is whether Tesla can build, approve, deploy, clean, charge, and operate a fleet at useful scale. Manufacturing is now part of the autonomy test.
|
|
Tesla and SpaceX keep erasing the line between them
On Tesla's earnings call, Musk pointed to growing overlap between the two companies. He did not confirm a merger. He did not close the door either.
The overlap is already visible. Tesla knows batteries, factories, power electronics, vehicles, and real-world AI. SpaceX knows rockets, global networks, extreme hardware, and high-rate production. Both now depend on huge compute and custom chips.
Why this matters: A legal merger is only one form of integration. Shared engineers, shared suppliers, shared silicon, and shared infrastructure can create many of the same gains. The empire can behave like one machine before the paperwork says it is one.
|
|
Starship's thirteenth flight gave SpaceX two very different lessons
The upper stage came through reentry with much less visible heat-shield damage. That is the hard part SpaceX must solve before Starship can fly like an aircraft instead of a one-use test vehicle.
The booster had the opposite ending. Five engines failed to relight for the final slowdown, leading to a harder impact than planned. One flight moved one side forward and exposed another weak point.
Why this matters: Reusable launch is a chain. Heat shield. Engines. guidance. landing. turnaround. One broken link can erase the gain from the others. SpaceX is getting closer because the failures are becoming more specific.
|
|
Musk is calling for direct safety talks between AI labs
In a recent interview, Musk argued that the leading AI labs should talk directly about safety and review major new systems. That idea lands at an odd moment. The same labs are racing harder than ever.
Safety review means testing a model for dangerous behavior before broad release. It does not require the companies to stop competing. Airlines compete and still share crash data. Chip firms compete and still follow common electrical rules.
Why this matters: Release speed raises the cost of silence. A fast lab needs fast ways to report serious failures. The mature version of the AI race is not slower. It is better instrumented.
|
THE MAIN EVENT
|
THE REAL PRODUCT IS THE LOOP
Model. Machine. Network. Repeat.
I keep seeing people judge each Musk company as a separate bet.
Is Grok smarter? Is Cybercab ready? Did Starship land? Did Tesla hit the margin target?
Those questions matter. But they miss the structure.
Grok is the model. Tesla is the machine. Starlink is the network. SpaceX is the transport layer. The factories turn software into atoms. The cars, robots, satellites, and users send reality back into the model.
That creates a loop.
A better model can improve code, factory planning, route decisions, and robot control. Better machines create more data. Better networks move that data faster. Better rockets expand where the network and machines can operate.
Then the loop starts again.
This is why release cadence matters more than one benchmark. A benchmark is a photograph. The loop is a film.
Grok 4.6 can fail. Cybercab production can move slowly. Starship can miss another landing. None of that breaks the thesis by itself.
The thesis breaks when the loop stops learning.
My take: Musk's deepest advantage is not ownership of many companies. It is the chance to make those companies teach each other. Today’s flood of launches, model numbers, and factory moves is what that learning loop looks like from the outside.
|
ELON ON X
|
MODEL CLOCK
“Grok 4.6 and 4.7” are already in the conversation.
The interesting part is not the decimal point. It is the message to builders: do not design around a model that stays still. Design around a stream of upgrades.
|
|
THE ODYSSEY BET
Musk says Grok Imagine can make a full-length Odyssey film this year.
That is a wild deadline. It is also a clean test. Long films need stable characters, scenes, voices, pacing, and memory. A three-minute clip can hide flaws. Two hours cannot.
|
|
MONEY AFTER ABUNDANCE
Musk hinted he may make a Buffett-style giving pledge.
He tied the idea to a future where AI and robots make goods abundant. The promise is vague. The tension is not. If money matters less later, what should the richest person do with it now?
|
THE TROLL REPORT
|
🧌 “Can we finish testing Grok 4.5 first?”
No. The decimal points have unionized. They now arrive in pairs and demand their own launch event.
|
|
🧌 “Tesla and SpaceX are totally separate.”
Of course. One builds robots, batteries, chips, and AI cars. The other builds robots, batteries, chips, AI networks, and rockets. Completely different family dinner.
|
|
“As you can tell from the many collaborations on so many fronts with SpaceX, there’s more and more overlap.”
— Elon Musk, Tesla Q2 2026 earnings call, via Reuters
|
|
The loop never clocks out.
Forward this issue to the person still judging the empire one company at a time.
READ ÆLON DAILY
|
Sources: X trend: Grok 4.6 and 4.7 · xAI: Grok 4.5 · Reuters: Tesla–SpaceX overlap · Reuters: Starship Flight 13 · Tesla Investor Relations · Forbes Real-Time Billionaires. Figures and product claims can change after publication.
|
