Return on Hype

We live in a world where a lean 10-person team can hit $100M ARR in 12 months, and 2 year old companies with minimal revenue grab billion-dollar valuations.

Some observations on my radar:

Small teams, outsized impact. A growing trend of small teams generating outsized revenue in a short amount of time. For instance, the AI IDE Cursor, with a team of 10, hitting $100M in 12 months. Eric Stromberg charmingly calls it “bootscaling”

Investor hype around AI remains at a feverish high. Perplexity tripled its valuation to $9B in six months. Figure AI is reportedly now worth 15x more from a year ago, at $39.5B. Looks like the Deep Seek impact on monster valuations didn’t last the week.

Microsoft’s Satya Nadella seems “done with the bubble”. AI and AGI is overhyped. Models are commoditized. Are we seeing the return on infrastructure buildout?

“Claiming some AGI milestone is some nonsensical benchmark hacking to me. The real benchmark to me — is the world [and its value] growing at 10%?”

I am inclined to agree.

Must-Know News This Week

Elon Musk’s xAI adds ‘Big Brain’ reasoning to Grok-3: Elon Musk’s xAI unveiled Grok-3, announcing that the new AI model has “more than 10 times” the compute power of its predecessor. There are two Grok-3 reasoning modes available: “Think”, which will display Grok’s reasoning as it resolves requests; and “Big Brain” for complex tasks that require more computational power. xAI is also launching a Grok AI agent product called Deep Search, which the company describes as a “next generation search engine.”

Thinking Machines Lab is ex-OpenAI CTO Mira Murati’s new startup: Former OpenAI CTO Mira Murati has announced her new startup. Called Thinking Machines Lab, the startup, which came out of stealth, intends to build tooling to “make AI work for [people’s] unique needs and goals,” and to create AI systems that are “more widely understood, customizable, and generally capable” than those currently available. Murati is heading up as CEO. OpenAI co-founder John Schulman is the company’s chief scientist, and Barret Zoph, OpenAI’s ex-chief research officer, is the CTO.

’Personalized, unrestricted’ AI lab Nous Research launches first toggle-on reasoning model: Nous Research launched a new open reasoning model called “DeepHermes-3 Preview,”. The model is described as an “LLM that unifies reasoning and intuitive language model capabilities,” and allows the user to switch at will between longer reasoning processes and shorter, faster, less computationally demanding responses.

Microsoft’s Xbox AI era starts with a model that can generate gameplay: Microsoft Research and Xbox game studio Ninja Theory have partnered to create a new AI model for gaming. Muse AI is a first-of-its-kind generative AI model that can generate a game environment based on visuals or players’ controller actions. It understands a 3D game world and game physics and can react to how players interact with a game. The Muse model was trained on a large amount of human gameplay data from the Xbox game Bleeding Edge. The model is limited to generating gameplay visuals at a resolution of just 300×180 pixels.

OpenAI cofounder Ilya Sutskever’s new AI startup is fundraising with a $30 billion valuation: Ilya Sutskever’s Safe Superintelligence (SSI) is in the process of raising in excess of $1 billion with a valuation topping $30 billion. Bloomberg reports San Francisco-based Greenoaks Capital Partners is leading the deal and plans to invest $500 million itself. The new figure is significantly higher than SSI’s $5 billion valuation in its last round, held this past September, when it raised $1 billion from investors including Sequoia Capital and Andreessen Horowitz. Very little is known about the company so far, aside from its stated goal of building a “safe superintelligent” AI system. The company does not yet have a product on the market.

Perplexity has its own ‘Deep Research’ tool now too: Perplexity introduced a new tool called Deep Research that it says can conduct “in-depth research and analysis” to deliver detailed reports in response to questions, and it’s free for limited use. It comes just a couple of weeks after OpenAI announced its own Deep Research feature for ChatGPT Pro users..which itself followed Google’s December announcement of Deep Research for Gemini. The company claims it outperforms competitors — like OpenAI’s o3-mini and o1, and DeepSeek-R1 — on the Humanity’s Last Exam benchmark, earning a 21.1% accuracy score (though this is lower than OpenAI’s Deep Research score).

Meta unveils Project Waterworth, a 50,000km subsea cable: Meta has announced Project Waterworth, its most extensive subsea cable initiative to date. Spanning over 50,000 km (31,000 miles), the system will connect five continents, including the U.S., India, Brazil, and South Africa. Once completed, it will be the world’s longest subsea cable network, designed to enhance digital connectivity and support AI-driven applications.

xAI reportedly seeks to raise $10bn for AI expansion: xAI is reportedly seeking a $10bn funding round, valuing the firm at approximately $75bn. Existing investors such as Sequoia Capital, Andreessen Horowitz, and Valor Equity Partners are in discussions to participate, reported Bloomberg, citing a source. The terms of the funding round are not yet finalised and could change.

Microsoft unveils quantum chip powered by a ‘new state of matter’: Microsoft unveiled Majorana 1, a quantum chip powered by an exotic state of matter that's neither liquid, solid, nor gas, called topological superconductivity. The breakthrough, detailed in the scientific journal Nature, uses a custom-built material known as a topoconductor to create and control Majorana particles, potentially scaling quantum computing to solve previously impossible problems. Current versions of the chip contain eight topological qubits, but Microsoft claims the architecture could scale to one million qubits on a single palm-sized chip.

Anthropic prepares new Claude hybrid LLMs with reasoning capability: Anthropic is getting ready to release a new AI model that combines traditional language model capabilities with advanced reasoning functions. According to The Information, the model will be available in the coming weeks, with a focus on enterprise applications. A key feature of Anthropic's new model is its variable resource allocation – users can adjust how much computing power the model uses for each task through a simple slider.

Figure AI to grab $1.5B funding at $39.5B valuation; eyes to produce 100,000 robots: Figure AI is in talks of securing a $1.5 billion in funding, increasing its valuation to $39.5 billion. This is 15 times more than its current valuation of $2.6 billion that it attained last year on raising $675 million. As per Bloomberg, Align Ventures and Parkway Venture Capital, both renowned for their strategic investments in cutting-edge technologies are said to lead this round. Notably, Bezos has amplified his commitment by pledging an additional $100 million through his venture firm, Explore Investments. Figure AI has unveiled plans to produce and deploy 100,000 humanoid robots over the forthcoming years.

Mistral announces its first regional language model – Mistral Saba, a 24B parameter model trained on meticulously curated datasets from across the Middle East and South Asia. Mistral Saba supports Arabic and many Indian-origin languages, and is particularly strong in South Indian-origin languages such as Tamil and Malayalam.

HP has acquired Humane's assets for $116 million, but will not continue the AI Pin hardware. The deal will integrate Humane's AI capabilities into HP's broader portfolio.