Contrary Research Rundown #122

Rumors of the death of X have been greatly exaggerated, plus new memos on Decagon, Metronome, and more

Feb 08, 2025

The Evolution of X

In October 2022, Elon Musk acquired Twitter for $44 billion. Sometimes it feels like all anyone has talked about since. As part of the acquisition, Musk raised $13 billion of debt to finance the deal. Since then, it had been hailed as “the worst buyout for banks since the financial crisis.” That is, until this week.

In January 2025, banks that had held onto the Twitter buyout debt since 2022 went out to try and tentatively sell $3 billion of it at a discount. The outcome surprised even them — they ended up selling $5.5 billion at just a 5% discount. Investors are clearly seeing financial viability in X’s future.

How did the worst buyout since the financial crisis manage such a minimal discount? The story is one you might think you’ve heard already, but the details would surprise you. Here’s how Elon Musk managed to turn the money-losing bird app into the latest entry in the Muskonomy.

Step one? Shed some feathers.

In July 2023, Elon Musk rebranded Twitter into X — the everything app. The stated goal of X was to build a single platform across social, financial, and media use cases. A pretty hefty vision.

Step two? Alienate your largest source of revenue.

In November 2023, Musk disbanded the company’s “Influence Council”, a select group of top advertisers on the platform. As advertisers started to express their concerns about spending money on the platform, pushing for content changes, Musk responded in no uncertain terms:

“I hope [advertisers] stop. Don’t advertise […] if someone’s going to try to blackmail me with advertising, with money, go f**k yourself.”

Step three? Emphasize who is really important — users.

Over the course of a few years, X launched more products than Twitter had in the prior decade. Creator subscriptions, revenue sharing with creators, a hiring platform, ecommerce shopping, and a P2P payments wallet.

From the beginning, Elon Musk’s vision for what X was capable of was building out from a core responsibility as the “de facto public town square” to a sweeping vision of becoming the “everything app.”

Despite dozens of high profile departures from X, users voted with their feet. In 2022, when Musk acquired Twitter, it had 368 million MAUs. By January 2025, X had grown to 611 million MAUs. Plenty of the new products on the everything app are nascent, but the vision is clearly compelling to the people who can’t stop using the platform.

Step four? Engage the “Muskonomy.”

Elon Musk has pitched the potential benefits that can come from the synergy between his various companies — Tesla, SpaceX, The Boring Company, Neuralink, X, and xAI. For X, that enabled the further expansion of the company’s product.

Founded in July 2023, xAI launched into the AI race with the intention of “understanding the true nature of the universe.” In November 2023, xAI unveiled its AI model, Grok, that was integrated directly into X and made available to X’s Premium+ subscribers. In December 2024, Grok was offered for free to all X users.

The synergies extended to help X’s financial health as well. In the past, Twitter had been used heavily for AI training data by other companies, often without getting paid for it. In July 2023, after Musk introduced usage limits on processing Twitter data pre-rebrand, he pointed to the need to prevent AI companies from scraping Twitter data for training purposes.

Since then, X has started monetizing access to its data. For example, one report indicated that xAI has paid X “hundreds of millions of dollars” to X for access to training data. This could open up other possibilities for monetizing access to that data from other AI companies.

Step five? Get ready to say “I told you so.”

As X continued to grow users and draw attention, advertisers were unable to stay away. Prominent advertisers, like Comcast, Disney, IBM, and Amazon, started to advertise on the platform again.

Musk’s adversarial relationship with advertisers has been around values, not strategy. But advertising on X isn’t going away. Musk’s key hires at the company have made sure of that. Executives like Linda Yaccarino, Angela Zepeda, and Yale Cohen, respectively former heads of marketing at NBC, Hyundai, and Publicis, have looked to smooth relationships with advertising.

Step six? Tighten up the ship.

Despite an expanded product suite and increased willingness of advertisers to work with X, the clearest picture of X’s reinvention is in its financial picture. Prior to Musk’s acquisition, Twitter had generated $5 billion in revenue and $682 million in EBITDA in 2021. Since, the company has seen revenue decline to $2.7 billion in 2024, but generated $1.3 billion in EBITDA, nearly 2x the profit.

Where do we go from here?

Elon Musk, and X as a platform, continue to attract the ire of a variety of different groups. But the evolution of X from an inconsistently profitable microblogging site to a broader platform is slowly but surely moving its way towards success. From a widespread migration away from the platform to the inevitable bankruptcy, the bear case for X has been proven false time and time again.

But what about the bull case? What if we dream the dream? What could happen if X ends up being a successful experiment? It speaks volumes to what can be done with leaner operations, it demonstrates the power of a platform approach to building product, and it proves that things don’t have to simply stay the same because “that’s how it’s always been.” The products we use, the brands we dedicate our time and attention to — they can evolve and improve.

If there is one takeaway from Twitter’s evolution into X, it is that dreaming the dream can, in some cases, become the reality.

For the full rundown on X’s evolution as a business, you can read our full 10K+ word report here!

Decagon offers an AI agent designed specifically for enterprise customer support that possesses "human-like" intelligence. To learn more, read our full memo here and check out some open roles below:

Research Engineer - San Francisco, CA
Brand Designer - San Francisco, CA

Sword Health is a digital health company offering virtual physical therapy for musculoskeletal (MSK) care. To learn more, read our full memo here and check out some open roles below:

Senior DevOps Engineer - Portugal-based Remote/Hybrid
Senior Fullstack Engineer - Portugal-based Remote/Hybrid

Metronome provides a usage-based billing platform for companies that seek to deploy a pricing model, precisely measure how much end customers use their products, and apply pricing changes. To learn more, read our full memo here and check out some open roles below:

Product Manager (Platform) - New York City, San Francisco Bay Area, or Remote
Software Engineer (Infrastructure) - New York City, San Francisco Bay Area, or Remote

Check out some standout roles from this week.

Parafin | San Francisco, CA - Analytics Engineer, Data Scientist, Product Designer (Experienced), Senior Software Engineer (Infrastructure), Integration Engineer, Product Manager (Capital Growth - Experienced)
Ascend | Palo Alto, CA, Singapore, or Remote - Backend Engineer, Field Data Engineer (APAC), Developer Relations Advocate
Cerebras | Sunnyvale, CA or Remote - Generative AI Inference Solutions Architect, Design Automation Engineer, Principal Product Manager (ML), AI Infrastructure Test Engineer

OpenAI co-founder Ilya Sutskever's artificial intelligence startup Safe Superintelligence (SSI) is in talks to raise funding at a valuation of at least $20 billion, quadrupling its previous $5 billion valuation from last September.
Shein is set to cut its valuation to around $50 billion for its potential London IPO, down nearly a quarter from its 2023 fundraising value, due to growing headwinds.
The researchers behind the s1 AI model were able to closely replicate OpenAI's multi-million-dollar o1 reasoning model for under $50 in cloud compute credits.
Tristan Tao published an article comparing the AI capabilities in four major business intelligence (BI) platforms - Looker, Power BI, Tableau, and HyperArc - focusing on their flexibility, accuracy, responsiveness, ease of use, contextualization, and explainability.
GM is laying off roughly 50% of the employees at its discontinued Cruise robotaxi business after deciding to no longer fund the unit.
Banking startup Mercury is in talks to raise ~$250 million after hitting $500 million in ARR and as investors are becoming more interested in fintech companies.
dbt Labs, an analytics engineering company, has surpassed $100 million in ARR and 5K customers, driven by significant adoption from Fortune 500 companies.
The rise of reasoning-heavy AI models with significant operating expenses at scale is challenging the traditional software valuation playbook, potentially making AI investing more attractive in the public markets than in private markets.
The U.S. government's and tech industry's focus on containing the spread of powerful AI technology through trade restrictions was misguided, as it was never going to be possible to fully contain this emergent technology.
DeepSeek could potentially disrupt the business model of Scale AI.
AI and new programming paradigms won’t kill software developers jobs but it will transform the role of programmers, who will need to acquire new skills to leverage these technologies and create more ambitious and valuable software applications.
Google has removed a pledge from its website that it would not use AI for weapons or surveillance, indicating a shift in its AI principles.
Figure has decided to drop its partnership with OpenAI and instead focus on developing its own in-house AI models for its humanoid robots, citing a "major breakthrough" that will deliver "something no one has ever seen on a humanoid" in the next 30 days.
Glean, a work AI platform, has achieved $100 million in ARR in just three years, reflecting the surging demand for AI-powered workplace intelligence.
Vercel is overhauling its serverless infrastructure to better support AI applications, which have different performance requirements than traditional web apps, with the introduction of its new "Fluid Compute" architecture.
Capital may matter more than ever after the development of advanced AI that can replace human labor, leading to a world where those with significant capital have a permanent advantage and the ability of money to buy results in the real world dramatically increases.
ChatGPT's subscriber base nearly tripled to 15.5 million in 2024.

At Contrary Research, we’ve built the best starting place to understand private tech companies. We can't do it alone, nor would we want to. We focus on bringing together a variety of different perspectives.

That's why applications are open for our Research Fellowship. In the past, we've worked with software engineers, product managers, investors, and more. If you're interested in researching and writing about tech companies, apply here!