- Charbel X
- Posts
- Meta's Throwing $65B At AI, A 1-Minute Color Combo Class and more
Meta's Throwing $65B At AI, A 1-Minute Color Combo Class and more
A venture as great as Manhattan
Good morning from Sydney,
Investment in AI is nothing less than extraordinary at the moment. From the USD $500B announcement last week, to Meta’s $65B commitment to AI investment and now DeepSeek upending OpenAI’s scores, this year will be insane.
As always, so much in today’s newsletter.
Have a wonderful week 🙏🏽
Yours in Wonder,
Charbel
Founder of Velvet Onion, Faster Zebra and more to come …
Today’s Highlights
AI: Meta’s $65 Billion AI Extravaganza: A Manhattan-Size Ambition
Design: Compatible Colour Combos for Graphic Design: What Goes Exceptionally Well with What
Science & Tech: A 3D-Printed Material 5x Stronger than Titanium
Founding: Create Clear Offers: Vagueness Shoves the Deal Towards “NO”
Product: Lean Graph Theory for Scaling Smarts
Today’s AI image: Imagining A 3D-Printed Airplane: 5x Stronger
Quote for the day: The Kind of People You’d Want to Avoid
AI
Meta’s $65 Billion AI Extravaganza: A Manhattan-Size Ambition
Meta’s CEO, Mark Zuckerberg, has unveiled a colossal $60–65 billion budget for 2025, with a laser focus on AI infrastructure.
The ambitious plan aims to crown Meta AI as the go-to assistant and establish Llama 4 as the industry’s cutting-edge model.
All You Need To Know
Meta’s preparing to unleash 1GW of compute power. Enough to house a data centre that could swallow a chunk of Manhattan.
By 2025, they’re aiming to stockpile over 1.3 million GPUs, creating one of the planet’s heftiest AI hardware arsenals.
The 2025 budget is a bold 70% jump from last year, showing Meta’s all-in on the AI gold rush.
Zuckerberg’s betting big on Meta AI, predicting it’ll charm one billion users by year’s end.
The announcement lands amidst a heated AI showdown, with OpenAI’s Stargate Project and DeepSeek R1 grabbing headlines.
Why is this a big deal?
The AI arms race is reaching fever pitch.
Meta’s monumental investment isn’t just about bragging rights. It’s a strategic move to outpace rivals like OpenAI, which is pumping $500 billion into its own infrastructure, and DeepSeek, whose R1 model has proven it can compete with the best at a fraction of the cost.
This level of spending reflects the stakes: whoever dominates AI infrastructure and tools will likely control the next era of tech innovation.
Also in AI
Design
Compatible Colour Combos for Graphic Design: What Goes Exceptionally Well with What
Choosing and blending colours in design can feel like a puzzle without a guide, but it’s a skill worth mastering. Let’s try and do the same.
The Colour Wheel: Your Creative Compass
It features:
Primary Colours: Red, yellow, blue—these are your base.
Secondary Colours: Orange, green, purple—formed by mixing two primaries.
Tertiary Colours: Hybrids like red-orange or blue-green, created by blending a primary with a secondary.
Understanding the Basics
Hue: The colour itself (e.g., red or green).
Saturation: How vivid or muted the colour appears.
Value: The lightness or darkness of a colour, adjusted with white or black.
Colour Harmony: Making Combinations Work
Use these tried-and-tested approaches for harmonious designs:
Complementary: Colours opposite on the wheel (e.g., blue and orange) for bold contrasts.
Analogous: Neighbours on the wheel (e.g., blue, blue-green, green) for a soothing palette.
Triadic: Three evenly spaced colours (e.g., red, yellow, blue) for balanced vibrancy.
The Psychology of Colour
Each colour tells a story:
Red: Passion, urgency, energy.
Blue: Trust, calm, professionalism.
Green: Growth, nature, health.
Yellow: Optimism, creativity, joy.
Black: Elegance, power, sophistication.
Ratios for Balance: Numeric Charms
60-30-10 Rule: Use 60% as your main colour, 30% as a secondary, and 10% for accents.
60-40 Rule: For two colours, let one dominate (60%) while the other complements (40%).
By understanding the principles of colour theory, you can design with confidence and create visuals that leave a lasting impression.
Now grab that colour wheel and start experimenting.
Also in Design
Pentagram’s Cyanotype Creativity: The Grow to Know features stunning cyanotype plant photography
Tesla’s Model Y Makeover: Sleeker lights, acoustic glass, a rear touchscreen, and suspension upgrades
Vercel Bags Tremor, Goes Open Source: Vercel has acquired Tremor, an open-source React library with 35 components for dashboards
Drafting in Space: UX for the Cosmos
AI Design Trends to Watch in 2025
Supergirl’s Logo Gets Gritty: Dark, grungy vibe might signal a moody storyline
Science & Tech
A 3D-Printed Material 5x Stronger than Titanium
The University of Toronto’s engineering researchers have created nano-architected materials that are as strong as carbon steel but as light as foam.
Combining machine learning with 3D printing, this breakthrough has the potential to revolutionise industries like aerospace and automotive.
How It Works
Nano-architected materials use carbon-based nanolattices for exceptional strength-to-weight ratios.
A machine learning algorithm optimised material designs, improving strength and reducing stress concentrations.
3D printing at the micro-nano scale enabled the creation of prototypes with double the strength of previous designs.
Global collaboration with institutions like KAIST, MIT, and Rice University brought diverse expertise to the project.
Replacing titanium with these materials in aerospace could save up to 80 litres of fuel per kilogram annually.
Why is this a big deal?
This innovation could transform industries by creating materials that are stronger, lighter, and more efficient.
The integration of machine learning in material design is a game-changer, offering smarter and more sustainable engineering solutions.
The potential applications are vast, from eco-friendly transportation to cutting-edge medical devices.
Also in Science & Tech
DeepSeek R1: A $5.6M masterpiece outperforming pricier models like OpenAI’s o1
DeepSeek’s Reinforcement Revelation: DeepSeek’s R1 leverages reinforcement learning over fine-tuning, excelling in coding and mathematics
AI Datacenters: Bigger Isn’t Always Better
Mom’s Microbiome Magic: New study links maternal gut health to fetal stem cell programming
Space Blurs Vision: Astronauts face temporary eyesight issues after extended microgravity exposure
Founding
Create Clear Offers: Vagueness Shoves the Deal Towards “NO”
Casting a wide net with your messaging might feel like a safe bet, but it often backfires.
A vague tagline like “I help leaders lead with impact and drive success” leaves too much to interpretation.
Without clear specifics, potential clients can't see themselves in your offer, and you risk losing them entirely.
Why be specific?
Specificity transforms your offer from a forgettable blur into a laser-focused solution.
When Justin Welsh refined his tagline from “I help SaaS companies grow their revenue” to “I help early-stage SaaS companies in healthcare scale from $1M to $10M with proven growth playbooks,” he not only narrowed his audience but also improved his success rate.
The clarity allowed ideal clients to self-select, saving time and boosting conversions.
Two Questions to Sharpen Your Offer
To avoid ambiguity, ask yourself:
Who is this for? Define your audience with precision—whether it’s middle managers, authors, or CEOs.
What will they achieve? Spell out tangible outcomes instead of abstract promises like “success” or “impact.”
When you narrow your focus, you’re not excluding people; you’re magnetising the right ones.
A targeted offer ensures that your perfect client feels like you’re speaking directly to them, while others can quickly self-disqualify.
Remember, you don’t need to appeal to everyone. Just the right ones.
Also in Founding
Lessons from Fin’s AI assistant flop: Users blame you for everything when you’re the middleman
Salesforce’s Sales Slip-Up: Even the software giant’s sales emails are a mess in 2025—bad lists, poor training, and cringe-worthy formatting
The Art of Doing Less: Quiet time isn’t lazy—it’s luxury
Asimov’s Take on Creativity: Unexpected connections and solo brainstorming
Product
Lean Graph Theory for Scaling Smarts
Lean principles—like limiting work-in-progress, continuous improvement, and systems thinking—can benefit any organisation but must be adapted to the specific environment.
Rapid-growth companies often dismiss Lean as "just for manufacturing," while large enterprises tend to over-focus on certain aspects, chasing diminishing returns.
3 Models for Organisational Dynamics
Path Graphs: Ideal for linear processes like assembly lines, task pipelines, and customer journeys with fixed stages.
Directed Acyclic Graphs (DAGs): Useful for branching workflows, approval processes, and goal cascades.
Network Graphs: Best for modelling collaboration, feedback loops, and ecosystem mapping.
Real-world organisations often blend these models to manage complexity.
It’s all about orchestrating the chaos with the right fusion of Lean principles and dynamic models.
I believe that even the messiest of networks can lead to very elegant solutions.
Today’s AI Image
Imagining A 3D-Printed Airplane: 5x Stronger
Quote of the Day
The Kind of People You’d Want to Avoid
"Don’t ever wrestle with a pig. You’ll both get dirty, but the pig will enjoy it."
Cale Yarborough
What we’re working on
Velvet Onion & Friends The new Velvet Onion & Friends will be launched soon. It’s our latest evolution, helping companies build products. It’s more than services. | Faster Zebra February 2025 - the product and venture school journey begins. Whitepaper launching in January. |