DEV Community: Anirudh Konidala

SpaceX + Cursor: The $60B Bet that Redefines Vibe Coding and the AI Stack

Anirudh Konidala — Wed, 22 Apr 2026 15:29:20 +0000

SpaceX builds rockets, launches satellites, and is working to make humanity multiplanetary. They also plan to go IPO very soon 👀 👀

Yesterday, they just did something that shocked the tech realm: a $60 billion bet on an AI code editor

That contrast isn't ironic, but rather the whole point. It tells you exactly where the real leverage in the AI era actually lives.

This isn't a bet on a product, but rather a bet on how software will be written.

The Deal

Yesterday, it was reported that SpaceX has secured an option to acquire Cursor, the AI-native IDE built by Anysphere, for $60 billion later this year.¹

This isn't a straightforward acquisition announcement. SpaceX has two choices:

Pay $10 billion for the work they are already doing together: a partnership payout
Exercise the option and acquire Cursor outright for $60 billion later in 2026

Think of it as an expensive try-before-you-buy. SpaceX is already working with Cursor. The $10B isn't a deposit, but instead is a measure of what the collaboration itself is worth if they don't follow through on the full acquisition. The $60B is the price of owning the whole thing.

This structure, optionality with a high floor, shows us that SpaceX isn't playing around. They are deeply committed regardless of which path they choose. Obviously, you wouldn't pay $10 billion for a partnership you're not serious about.

Why This Deal Exists

To understand why SpaceX wants Cursor, you need to understand how the AI stack is structured.

Building a dominant AI company requires three things:

Compute → Models → Interface

Compute is the raw infrastructure. This includes GPUs, data centers, training clusters
Models are the intelligence layer trained on that compute
Interface is where users actually interact with the models, the product layer.

Most companies only own one of these

OpenAI has strong models but relies on Microsoft's compute and third-party interfaces
Hugging Face has distribution but no proprietary compute or models. Even the best-funded AI labs tend to be missing at least one leg of the stool.

SpaceX, after acquiring xAI back in February for $250 billion², now controls two legs:

Compute: Colossus, SpaceX's supercomputing cluster described as equivalent to one million H100 GPUs³
Models: xAI's Grok, and all the research and infrastructure that comes with it

What they don't control is the interface: the layer where developers actually work to spend their time

Cursor is what makes that layer happen for SpaceX

Cursor's Role

Cursor is not a plugin. It's not an autocomplete tool bolted onto an existing editor.

It is an AI-native IDE, a development environment rebuilt from the ground up as a fork of Visual Studio Code, with the assumption that AI is a first-class participant in the programming process. It handles multi-file context, runs background agents, integrates with the terminal, and supports workflows where the human is directing rather than typing.

By April 2026, Cursor is used by 70% of Fortune 1000 companies⁴, with 60% of its revenue coming from enterprise deployments. We're talking Nvidia (all 40,000 engineers), Salesforce (90% of ~20,000 developers), Uber, Adobe, Shopify, OpenAI, and Perplexity. Its ARR trajectory is the fastest ever recorded for a B2B software company:

Date	ARR
January 2025	$100M
June 2025	$500M
November 2025	$1B
February 2026	$2B

That growth isn't just about a good product. It reflects a structural reality: when developers adopt a tool, they adopt it deeply. Their muscle memory, their workflows, and their shortcuts all adapt to the tool. Switching costs are high, but retention is high. And the tool becomes the surface through which everything else reaches them.

Interface equals distribution. Whoever owns where developers work owns the channel to every capability built on top.

This is why Cursor is worth $60 billion to SpaceX. Not because of the editor itself, but because of what the editor unlocks for big businesses.

The Real Shift: From Vibe Coding to Agentic Development

"Vibe coding" entered the mainstream vocabulary in early 2025. It was coined by Andrej Karpathy, a founding member of OpenAI and former Director of AI at Tesla, where you could describe what you wanted in natural language and an AI would write the code. It was playful, accessible, and genuinely useful for prototyping.

But vibe coding was only the start. Since then, things have evolved like crazy:

<span>Autocomplete</span><span>→</span><span>Copilots</span><span>→</span><span>Agents</span>
<span>(line-by-line)</span><span></span><span>(file-level)</span><span></span><span>(task-level)</span>

Each step represents a shift in what the developer actually does. With autocomplete, you still write most of the code. With copilots, you accept or reject suggestions. With agents, you describe a goal and review the output.

The workflow has inverted. The developer is no longer primarily a writer of code, but rather they are an orchestrator of AI systems. They define requirements, review outputs, catch errors, redirect when the agent goes wrong. The cognitive work has shifted upward: less syntax, more architecture; less implementation, more judgment.

This shift has a concrete consequence: the interface layer becomes dramatically more important.

When developers were writing every line, the value of a tool was in how fast it helped you write. Now that agents handle more of the implementation, the value is in how well the tool helps you think, direct, and review. That's a different product entirely — and Cursor is further along that curve than anyone else.

The SpaceX / xAI Strategy

Put the pieces together and the thesis becomes clear.

SpaceX acquired xAI to control the model and compute layers. The Colossus supercomputer gives them training capacity that few organizations on Earth can match. Grok gives them a model family to improve and deploy.

But raw capability without distribution doesn't win markets. Google has arguably the best AI research lab in the world. It still doesn't dominate the developer tools space. Research excellence and product dominance are different problems.

Cursor solves the distribution problem. It gives SpaceX/xAI a direct channel to the engineers at 70% of the world's largest companies — engineers who are already using the tool daily, already trusting it with their codebases, already building workflows around it.

The long-term play writes itself: gradually integrate xAI models as the inference layer inside Cursor. Use Colossus to train models on the coding patterns flowing through the platform. Build a feedback loop where more usage generates better models, better models drive more usage.

This is vertical integration applied to AI infrastructure: compute → models → interface, all inside one company.

It's the same logic that made Apple powerful — own the hardware, the OS, and the app layer — applied to the AI developer stack.

The Most Valuable Layer Is the Interface

This is the core insight that makes the deal make sense.

In technology, the layer that touches users most directly tends to capture the most value over time. Not because it's the hardest to build, but because it shapes behavior. And behavior shapes everything else.

Developers follow tools. They learn keyboard shortcuts, build mental models, develop preferences. Over time, the tool doesn't just serve the workflow, but rather defines it. The tool becomes the lens through which developers understand what's possible

When that tool is also the delivery mechanism for AI models, the stakes compound. Cursor doesn't just influence how developers write code today. It shapes what kinds of AI capabilities they discover, adopt, and come to depend on. It determines which models feel natural and which feel foreign. It becomes the default.

The company that owns the default developer interface owns the on-ramp to the entire AI stack.

That's worth $60 billion, maybe more

Second-Order Effects

If this deal closes, or even if the $10B partnership proceeds, several structural shifts will most certainly follow

IDEs become strategic assets - Until recently, development environments were productivity tools. Now they're territory. Every major AI lab and cloud provider will look at their developer tooling and ask whether they own their on-ramp or rent it from someone else. Expect consolidation.

AI labs will move up the stack - The lesson of this deal is that model quality alone doesn't guarantee distribution. Labs that want to avoid becoming commodity inference providers need a product layer. Anthropic has Claude Code. Google has Gemini Code Assist. The race to own the interface is now explicit.

The developer role continues to shift - As agentic workflows mature, the skills that matter most change. Developers who thrive will be those who are good at directing AI systems — writing clear specifications, reviewing outputs critically, catching subtle errors. The demand for developers who can only write code mechanically will compress; the demand for developers who can architect and judge will grow

Vendor lock-in gets embedded deeper - When your IDE is also your model provider, switching costs multiply. Changing your AI coding tool no longer means just changing a plugin, but rather involves re-learning workflows, losing context, and migrating integrations. This is good for Cursor's retention and potentially bad for developer autonomy

Risks and Tradeoffs

This deal also introduces real concerns worth naming.

Model bias - If Cursor defaults to xAI's Grok, developers using the tool will have their defaults set by a company with its own interests. The model that autocompletes your code, drafts your documentation, and reviews your pull requests is not neutral. Whoever controls that model controls a subtle but pervasive influence over developer output

Centralization of critical infrastructure - Software development is already highly concentrated around a few tools and platforms. A world where a single company controls the compute, the models, and the primary developer interface is one with very little redundancy. That's a systemic fragility, not just a competitive concern

Lock-in at the workflow level - As developers build habits around Cursor's agentic features, the switching cost grows. This is rational from a business perspective. But it means the developer community becomes increasingly dependent on the decisions of one company: its pricing, its model choices, its policy on what the tool can and cannot do

These aren't reasons to dismiss the deal. They're reasons to watch it carefully

Conclusion

Software development is in the middle of a structural transition. The question is no longer just "how do we write code faster?" It's "who owns the infrastructure through which intelligence is directed?"

SpaceX's bet on Cursor is an answer to that question, with a price sticker towards $60 billion. Getting there as one of the first companies to do so is worth even more, especially for a company that is going to go public very soon!

The shift underway is real: developers are moving from writing code to directing intelligence and t interface where that direction happens is not a peripheral concern, but rather the central strategic asset of today's AI era.

The companies that understand this are moving fast, building fast, and shipping fast. The ones that don't will find themselves building excellent models that reach developers only through someone else's tool and on someone else's terms.

The rocket company bought a code editor. That tells you everything about where the leverage is right now in today's AI era

References

OpenClaw + NemoClaw: How GTC 2026 is already the most important week in Agentic AI

Anirudh Konidala — Wed, 18 Mar 2026 04:17:01 +0000

The age of Agentic AI is no longer coming: it's here. And NVIDIA just built the entire stack for it

That's what NVIDIA's CEO, Jensen Huang, said while walking into the SAP center during NVIDIA's 2026 GTC conference in San Jose in front of thousands of engineers

GTC 2026 wasn't just a product event: it was a declaration

NVIDIA's GPU Technology Conference (GTC) is the techbro and AI world mere equivalent of the State of the Union address. 450+ sponsors, 1,000 sessions, 2,000 speakers, and a CEO who has surprisingly been right about where AI and computing, in general, is going

NVIDIA's GTC 2026 runs all the way through March 19th, with more sessions and announcements still to come, but there is already so much information that I am eager to talk about how it is changing the agentic AI landscape!

During this year's GTC, Huang summarizes that

the token is the new line of compute

In this era, the new benchmark that matters is tokens per second and tokens per dollar

That one statement tells you everything. It confirms that NVIDIA isn't just selling GPUs anymore. They are slowly building the economy of intelligent machines. And during this year's GTC, they proved exactly how agentic AI fits into that vision at every single layer of the stack

Here's how GTC 2026 reshapes the agentic AI landscape and why builders need to be paying close attention

The Unprecedented Signal of Demand

Huang opens up the conference with what seems like an exaggeration, albeit true: over the last few years, computing demand has increased by 1 million times. This was with all due respect towards and the plethora of AI-native companies: OpenAI, Anthropic, and hundreds and hundreds of AI startups that have altogether pulled in $150 billion in venture capital (VC) last year alone

Obviously, AI-native companies need GPU computing, specifically from NVIDIA: the unanimous leader in the GPU market. Therefore, Huang states that GPU demand is "off the charts"

NVIDIA now projects at least $1 trillion in revenue from 2025 to 2027, where one analyst considers NVIDIA as the king of compute inference

This is a signal: one where the infrastructure layout and buildup for agentic AI is being measured in trillions and the companies building on top of this stack are still very early

OpenClaw + NemoClaw: the Agentic OS the industry has been waiting for

The recent success of OpenClaw drew many developers. It was so successful Huang considered it as "the most popular open source project in the history of humanity". Sounds extreme, but when you have a deeper understanding of what it does, it makes sense

OpenClaw made computers' operating systems fully agentic. It cancan execute OS tasks via LLMs. It can send emails and Slack messages, host Microsoft Teams meetings, and complete workflows autonomously

Shipping an agent isn't the hardest part. The trust boundary is. An agent should be taking real-world actions in a production environment: calling APIs, processing payments, reading internal docs, and interacting with external services. However, AI agents need precise control over what they can do, when, and for whom

Without a real guardrail layer and security, they become autonomous systems with no seatbelts

Every builder working on production agents is reinventing some version of that security layer, but NVIDIA's NemoClaw completely blew this aspect away

NemoClaw is NVIDIA's enterprise stack built directly on top of OpenClaw. It is OpenClaw with a policy engine, a firewall, and audit trails baked directly into an OS's execution layer. Three pillars:

Policy Enforcement: You define what your agent can do and NemoClaw enforces it at runtime. Not at the prompt level, but deep in the execution layer of the operating system where it actually matters
Network Guardrails: Controlled exits/terminations for sensitive workloads. Agents can't exfiltrate data outside their scoped access. For any enterprise deployment where agents touch internal systems, this is a "safety hazard" for a production system
Privacy Routing: Proprietary data doesn't leak to third-party model providers. For companies running agents over internal knowledge bases, financial records, or customer data, this is what makes legal and compliance actually sign off

Huang states that NemoClaw will be the "policy engine of all the SaaS companies in the world". And every enterprise sitting on valuable data that wants to deploy agents needs exactly this

NVIDIA OpenShell Runtime

NVIDIA also announced the OpenShell runtime during this year's GTC. It is the execution environment that sits between your agent's decision-making and the infrastructure it runs on. OpenShell is what makes NemoClaw's policies enforceable at scale, along with the composable glue layer for deploying agents consistently across teams, services, and infrastructure boundaries

Previously, developers would have to

Pick a framework
Build their own authentication layer
Hand-roll policy enforcement
Wire up their own observability
Ship something possibly fragile

Now, with OpenClaw, NemoClaw, and OpenShell, developers can

Define their agent's tools and context
Configure policy in NemoClaw
Deploy via OpenShell
Focus entirely on the intelligence layer

Now enterprises can safely trust agents with this new agent tech stack, unlocking the next wave of enterprises to use AI Agents

The Nemotron Coalition: An Open Model for Every Domain

NVIDIA isn't just building a one-and-done model. They are building a ecosystem of open frontier models across every major domain

Model Family	Domain
Nemotron	Language & Reasoning
Cosmos	World & Vision
Isaac GR00T	General-Purpose Robotics
Alpaymayo	Autonomous Driving
BioNeMo	Biology & Chemistry
Earth-2	Weather & Climate

They keyword is open. Nemotron's set of models fine-tune on your domain data, deploy via NVIDIA's NIM microservices, skip the frontier API costs for tasks where a specialized smaller model wins on both performance and price. Pair this with NemoClaw's policy layer and this forms a complete, self-contained agentic stack, containing intelligence, guardrails, execution, and deployment

NVIDIA continues to reshape inference computing

Vera Rubin is NVIDIA's new flagship platform: 7 chips, 5 rack-scale systems, and one supercomputer built specifically for agentic AI workloads. It includes the new Vera CPU and BlueField-4 STX storage architecture. Microsoft Azure is already running Vera Rubin NVL72 systems globally!

Feynman is what comes next: a new CPU named Rosa (that is designed to move data, tools, and tokens across agentic infrastructure), the LP40 LPU for inference, BlueField-5 networking, and NVIDIA Kyber for co-packaged optics. This spells out every pillar of the AI factory, including compute, memory, storage, networking, and security. Each pillar just got a generational upgrade by NVIDIA themselves

NVIDIA also has plans for space with Space-1 Vera Rubin: AI data centers built for orbital deployment. NVIDIA took that as a opportunity to extend accelerated computing beyond the atmosphere entirely!

The Next Deployment Surface for Agents: The Physical World

Huang simply stated that

The ChatGPT moment of self-driving cars has arrived

Major car manufacturers like Nissan and Hyundai have joined NVIDIA's DRIVE Hyperion robotaxi platform and Uber is deploying these vehicles commercially into its ride-hailing network. T-Mobile is also evolving base stations into edge AI platforms for their physical AI workloads

Therefore, the next step for AI agents is the physical world: agents that see, reason, and act in real environments

NVIDIA Now Owns the Entire Agentic Stack

NVIDIA announced so much during this year's GTC conference

Intelligence layer: Nemotron Coalition open models
Agent OS: OpenClaw
Trust + Policy: NemoClaw + OpenShell that pairs with OpenClaw
Inference: NIM microservices, and positioning themselves to be the "computing inference king"
Compute: Vera Rubin → Feynman → Space
Physical: Cosmos, Isaac GR00T, DRIVE Hyperion

That's proves that NVIDIA has deliberately decided to own agentic AI, where every layer is covered and every boundary is addressed

This is what reshapes agentic AI. Not one announcement, but rather the whole, complete stack at once

Final Words

GTC 2026 wasn't a series of announcements. It was a map Jensen Huang laid out as to how NVIDIA intends to be the foundation of the age of Agentic AI and intelligent machines

For those of us building agents right now, the sky is the limit. The tools have never been this good and the platforms have never been this complete

The only remaining question is how the rest of us are going to make the most of them

The Story of OpenClaw: From 200K+ GitHub Stars to an OpenAI Product

Anirudh Konidala — Tue, 17 Feb 2026 12:14:52 +0000

Recently, Sam Altman (OpenAI's CEO) made an announcement on Twitter (X) that Peter Steinberger, the creator of the viral open-source agent, OpenClaw (formerly Clawdbot), will be joining OpenAI

// Detect dark theme var iframe = document.getElementById('tweet-2023150230905159801-154'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2023150230905159801&theme=dark" }

The big question for the community though is: What will happen to OpenClaw?

Fortunately, Altman confirmed that OpenAI won't kill this internet sensation of an AI agent (see X/Twitter post above). Rather, they will continue to keep it as open source and fully integrate it into the company, where Steinberger will be at the forefront towards accelerating the development of personal AI Agents for the company!

Why OpenClaw?

While many AI agents already existed, they could really only help with one specific task and couldn't handle the reality of the modern workday for the average adult: checking emails frantically, responding to Slack / Microsoft Teams messages, performing code reviews, and managing calendar invites all at once.

For example, imagine you are a developer. You might ask a traditional AI agent to summarize emails and it will do exactly that. However, while it's processing those emails for summarization, your Slack is blowing up with a critical production bug. The agent though is trapped in that one conversation of summarizing your emails. You’d have to wait for the email summary to finish, manually open a new tab, copy the bug report, and then ask the agent to help you debug it

This meant that users were still stuck halfway, and had to manually bridge and transition between different apps to handle tasks that were needed for a modern workday. acting as the manual bridge between different apps, which often felt like more work than just doing the tasks themselves.

OpenClaw solves this problem by giving you a single agent that works across all your tools simultaneously. It monitors your email, checks your Slack for any new messages, manages and works with your calendar invites, and helps conduct code reviews: all in parallel.

This makes it easy so that when a production bug hits while you're in the middle of an email summary, OpenClaw catches it, flags it, and helps you debug it without you ever having to switch tabs or start over.

Another hint that we are transitioning fully into AI Agents

For the past couple of years, AI was just a chatbot. You ask a question, and a chatbot would give you an answer

OpenClaw drastically changes that. It marks a shift towards agentic AI and AI agents! While a chatbot like ChatGPT gives you a draft of an email, an AI agent like OpenClaw will actually log in to your Microsoft Outlook account and sends the email!

It takes full control of your operating system and local files, turning the AI from what was once a research assistant into something that acts as a personal AI agent for your computer!

Its capabilities were so vast and immediately got many people hooked, and OpenClaw became one of the fastest growing open source projects, with now having over 200K+ stars on GitHub!

By bringing Steinberger on board, OpenAI is signaling that the era of the chatbot is over. We are moving from an AI that talks about work, to an AI that finishes it.

How Anthropic made the greatest fumble of all time

OpenClaw was previously known as Clawdbot, as a playful tribute and testament towards Anthropic's LLM Claude. Interestingly enough, Steinberger was so deeply invested into the Claude ecosystem that he made OpenClaw recommend Claude Opus 4.5 as the default LLM for its users.

But instead of seeing what could have been a partner, Anthropic saw it as a competitor, viewing Clawdbot as a trademark threat to their signature LLM: Claude. This prompted them to request Steinberger to change Clawdbot's name. Therefore, Steinberger ended up frantically changing the name twice from Clawdbot -> Moldbot -> OpenClaw

While Anthropic was worried about the name, OpenAI saw the true talent and potential with OpenClaw and immediately extended an offer to Steinberger, providing him the opportunity to build personal AI agents for everyone!

A product that was once a massive driver towards Anthropic API usage now became a product maintained by OpenAI! And what would have became a wonderful opportunity for Anthropic was now directly handed to their competitors at OpenAI!

The Lesson: Growth over Competition

Instead of viewing new projects as competition, developers and companies should view them through a growth lens.

The most successful moves that happen in tech aren't about vanquishing a rival, but rather about amplifying a vision through collaboration and/or acquisition.

For example, when Facebook acquired Instagram in 2012, they never saw it as a threat to Facebook. Rather, they saw the possibility of having social media in your pocket!

While many thought this $1 billion acquisition was an overstatement, Zuckerberg realized that Instagram perfectly captured the essence of social media in the mobile era, which Facebook was struggling to capture.

What does this mean for the rest of us

The success of OpenClaw isn't just about one developer landing a dream role at OpenAI. It signals two key learning takeaways we all can walk away with

AI is no longer living in your browser tab

As I mentioned earlier, we have fully moved away from talking to a chatbot in a browser tab. AI is no longer something you go to, but rather is now becoming embedded directly into your operating system by running in your computer's background as a daemon and acting on your behalf without you having to ask.

However, I believe the most important takeaway is that

Opportunities with companies like OpenAI, Meta, and Google don't always come from clicking "Apply" on LinkedIn or Workday

Steinberger didn't get a call from Sam Altman because of a polished resume or the right recruiter. He got it because he built something in public that 200,000 people found valuable enough to star on GitHub. He shipped obsessively, shared openly on GitHub, and solved a real problem that real people had.

The application button is one path, but it's the most crowded one. There are other paths to reach the same destination.

Steinberger showed that success can come by building things that speak for themselves: your GitHub, your open source contributions, the side project you've been putting off because it "isn't ready yet"

His story is proof that as AI is shifting from a chatbot to a multi-step agent (from talking about work to actually doing it), the developers who will shape that future aren't the ones waiting to be hired

They're the ones who are already building publicly

References

https://www.forbes.com/sites/ronschmelzer/2026/02/16/openai-hires-openclaw-creator-peter-steinberger-and-sets-up-foundation/

https://www.reuters.com/business/openclaw-founder-steinberger-joins-openai-open-source-bot-becomes-foundation-2026-02-15/

https://www.cnbc.com/2026/02/15/openclaw-creator-peter-steinberger-joining-openai-altman-says.html

https://www.businessinsider.com/sam-altman-hires-openclaw-creator-peter-steinberger-personal-ai-agents-2026-2

https://africa.businessinsider.com/news/clawdbot-creator-says-anthropic-was-really-nice-in-renaming-email-but-everything-went/50k0xe8

https://x.com/aakashgupta/status/2023249490426388849

SpaceX just made the largest acquisition in history. Here's why it matters.

Anirudh Konidala — Tue, 03 Feb 2026 20:28:35 +0000

The Elon Musk empire has only grown stronger

Yesterday, SpaceX acquired xAI for $250 billion, making it the largest acquisition of all time.

This acquisition will now value SpaceX at a $1.25 trillion dollar company¹

It only took xAI under 3 years to reach a $250 billion valuation. In this 3 year span, Musk merged X (formerly Twitter) into xAI, which then built its own LLM Grok on top of it, turning what was once a social platform into a full-on AI and Data engine

OpenAI took 9 years to reach that valuation. Anthropic took 5

With this acquisition, Musk isn't just consolidating his empire. He has a vision: one that is far ahead from the rest of us

That $1.25 trillion is a bet that AI's real bottleneck isn't LLMs. It's the infrastructure and data that it relies on²

xAI loses $1 billion a month competing with major LLM companies like OpenAI, Anthropic, and Google⁴

Now pair this with the challenging task of terrestrial data centers being unable to scale fast enough without crushing local power grids⁵ ⁶

This is where SpaceX saw an opportunity to solve AI's energy problem from orbit. And the first step involved merging xAI with SpaceX

SpaceX plans to utilize this acquisition to launch 1 million satellites and build AI data centers in space using solar power⁸

Musk claims that space-based AI will be the cheapest compute option in the next 2-3 years⁷

And the pieces have already been set in stone to prove Musk's claim. SpaceX launches the satellites, Starlink provides the connectivity, xAI provides the LLM's, and X provides the data. Each piece that belongs to Musk ends up feeding and satisfying one another.

Starlink's 9,000+ satellites in orbit already generate more revenue than SpaceX's launch business as a whole. With this recent acquisition, that cash flow will offset xAI's burn rate, while also giving SpaceX another reason to launch a million more satellites⁸

What this means for the current AI era

This acquisition isn't just another one of Musk's stories. It signals where the industry is heading

The AI race is no longer just about who has the best model, but rather who controls the infrastructure

OpenAI has Microsoft. $13 billion has already been invested by Microsoft so that Microsoft Azure can power every ChatGPT query
Amazon and Google have funded billions of dollars towards Anthropic, integrating Claude access into Amazon Web Services (AWS) and Google Cloud Platform (GCP)

Now, xAI is a part of SpaceX with its own unique plan to build its own infrastructure in space