Daily Digest: Claude and Pi

PLUS: Our profile of Autoblocks

Sign up | Advertise | Ben’s Bites News
Daily Digest #264

Hello folks, here’s what we have today;

PICKS

  1. Inflection AI’s Pi gets real-time web access. There’s no plugin/or additional switch. Just chat with Pi for recent info and Pi has it. Definitely the cleanest UI.

  2. Anthropic’s Cluade is now available in 95 countries. Its ability to chat with longer documents (up to 100k tokens) makes it stand out. Here’s the list of all supported countries.

  3. 📝Startup profile: Autoblocks - Our profile on how Haroon, Adam, and Hamza are building the platform for businesses to monitor LLMs apps. (preview below)

  4. Shaky economics of GPU reselling startups - These startups rent GPUs from providers and resell access at a markup to AI developers. But they compete fiercely for limited GPU supply and sell at slight premiums to cover costs, capping margins around 50%.🍿Our Summary (also below)

  5. Runway ML is launching an Acceleration Program to help exceptional software engineers become ML practitioners. It will be a 3-month program where you’ll be paid as a full-time employee at Runway working on their ongoing projects.

from our sponsor

Turn word-of-mouth into an MRR machine with Rewardful's affiliate tracking tool.

Loved by AI companies for its simplicity, perfect for SaaS using Stripe. Join the likes of Copy.ai, PDF.ai, Speechify & drive more MRR.

TOP TOOLS

  • Cosine - Your AI co-developer with semantic code search and code completion for your codebase.

  • Stable Audio Tools - Training and inference code for audio generation models from Stability AI.

  • Revision by Thiggle - Generative text to UI.

  • Expresso - Revolutionizing mental health in the workplace.

  • CapCut for Business - AI scripts and characters for your brand’s video ads at no extra cost.

  • NueronPedia - GeoGuesser for AI safety, no technical knowledge required.

WHO’S HIRING IN AI
NEWS
STARTUP PROFILE

Autoblocks is a platform helping businesses dealing with large-scale user data to better monitor their LLM-powered products. It allows engineering and product teams to debug error types, test code changes, and fine-tune datasets.

The company was founded in the spring of 2023 by brothers Haroon and Hamza Choudery, and Adam Nolte.

Together with their founding engineer, the Autoblocks team caters mostly to businesses looking to gain a competitive advantage by creating differentiated LLM-powered products.

The team

There are three founders—Haroon, Adam, and Hamza—along with a founding engineer. Haroon and Hamza specialise in growth and strategy, while Adam collaborates with the founding engineer on building the platform. Collectively, the team maintains a product-oriented focus.

From left to right: Hamza, Haroon and Adam

Values

Customers are the compass. Committed to customer obsession, they make every decision with their customers' success in mind. All team members engage with users, understand their needs, and innovate on their behalf.

Compete for existence every day. In a rapidly changing market, the Autoblocks team avoid complacency by striving to innovate and deliver value daily. Waiting for opportunities < actively seeking challenges to fuel growth.

Excellence in everything. Strive to exceed expectations in every aspect of the work. Excellence isn’t just a goal; it’s a habit.

Embrace a growth mindset, prioritising learning over ego. Abilities can be developed. Approach work with humility and a growth mindset. Take on challenges and find imaginative solutions.

Communication is a gift. Transparency and effective communication are the Autoblocks pillars. The team shares openly, provides clear information, and encourages stakeholder input to foster understanding and drive innovation.

QUICK BITES

The business models of GPU reselling startups have some underlying issues. These startups rent GPUs from providers and resell access at a markup to AI developers. But they compete fiercely for limited GPU supply and sell at slight premiums to cover costs, capping margins around 50%.

What is going on here?

GPU resellers are seeing rapid growth but have slim margins and other challenges that make their sky-high valuations questionable.

What does this mean?

Despite this, some like Together are being valued by VCs at 15x revenue, more typical of high-margin software companies. This lofty valuation assumes Together's value will continue rising 10x to reach $1B at the least.

Investors seem to be applying software-like multiples based on revenue growth, ignoring the fundamental challenges of GPU supply, low margins and potential saturation. But public markets may apply a lower multiple based on its thin margins, making that growth difficult.

Why should I care?

This disconnect between valuation and core business reality is problematic. It pressures startups to pursue unsustainable growth and raises the spectre of painful future downrounds. More reasonable valuations tied to actual margin potential rather than hype are needed.

Understanding the realities these startups face is important to assess their true potential. Their services fill a clear market need today, but margins, growth trajectories, and end markets matter. Beyond investors, founders and developers should be eyes-wide-open about the challenges these resellers face when looking for GPU vendors.

Ben’s Bites Insights

We have 2 databases that are updated daily which you can access by sharing Ben’s Bites using the link below;

  • All 10k+ links we’ve covered, easily filterable (1 referral)

  • 6k+ AI company funding rounds from Jan 2022, including investors, amounts, stage etc (3 referrals)

Reply

or to participate.