Most important AI news of the week 17th-24 August 2024 [Livestream]

AI Infra spend, Google wakes up and Meta Doubles Down on their App Thesis more

Aug 27, 2025

Thank to everyone for showing up the live-stream. Mark your calendars for 8 PM EST, Sundays, to make sure you catch them regularly.

Bring your moms and grandmoms into my cult.

Before you begin, here is your obligatory reminder to adopt my foster monkey Floop. He’s affectionate, relaxed and can adjust to other pets, kids, or people. No real reason not to adopt him. So if you’re around NYC, and want a very low maintenance but affectionate cat— then consider adopting him here.

Community Spotlight

Robert Wu

is a relatively new addition to my reading lists, but man has not missed at all. He drops some really good analysis on the Chinese Macroeconomic markets, and where they’re headed. Given how important China is becoming in AI, understanding the economic conditions driving growth is non-negotiable. Trying to focus only on the technical aspects while ignoring the larger socio-economic reasons is foolish, and Rob is one of the best sources for understanding the larger picture.

His most recent article: “Five structural forces that could extend China's "Quiet Bull" into a generational shift” is a great place to start if you want to know his work better.

If you’re doing interesting work and would like to be featured in the spotlight section, just drop your introduction in the comments/by reaching out to me. There are no rules- you could talk about a paper you’ve written, an interesting project you’ve worked on, some personal challenge you’re working on, ask me to promote your company/product, or anything else you consider important. The goal is to get to know you better, and possibly connect you with interesting people in our chocolate milk cult. No costs/obligations are attached.

Additional Recommendations (not in Livestream)

The Electric Slide by
Packy McCormick
is transcendently good. I won’t pretend to understand all of it, but Packy went above and beyond to go through the electric industry and tie it into where we need to proceed next. I usually don’t rank these lists, but this is by far the best read on the list.
“Beyond Formalism: A Rebuttal to Limits on LLM Reasoning” by
Interesting Engineering ++
is a fantastic read on the limits of reasoning and trends. Not sure if I agree that hybrid systems “reason”, but those are more semantics at this stage.
“Building Seemingly Conscious AI” by
Patrick McGuinness
really pushes a lot of interesting ideas in building the next generation of AI.
Austin Lyons
wrote an amazing article on Arms chiplet ecosystem. It’s a very important read for anyone and Austin is one the best analysts on this platform, so you should absolutely subscribe to him. I’ve never regretted reading an article, even when I have to spend a while understanding them.
Ben Dickson
has some of the best takes for AI on this platform. His “Boston Dynamics' boring new video is its most important yet” is no exception.
This is a great roundup of the major legally important AI developments by
Tobias Mark Jensen
“The Materials Behind the Machines” by
Meg McNulty
is a must read to understand how and why chips and other electronics are becoming strategic assests. It’s a great companion piece to
Packy McCormick
’s electric piece.
Cameron R. Wolfe, Ph.D.
went stupid in this overwhelmingly good look into GPT-oss. I don’t know what he puts in his water, but I need some of that.
Maryann
is one the best sources for understanding the inequality in health outcomes for women. Building AI means grappling with the uncomfortable reality that we might optimize a system with systemic inequality. Sources like her are important to build more empathy and a broader perspective required to not do that. Her “Why Viagra Was Fast-Tracked And Women’s Hormones Were Left Behind” is a must read for everyone.
Damien Benveniste
wrote “Mixture-of-Experts: Early Sparse MoE Prototypes in LLMs” an excellent deep dive into MoE.
“Scaling AI Governance in Healthcare” by
Sarah Gebauer, MD
more than delivers on the premise of this article.
“AI for Drug Discovery: Weekly News 🧢” by
Marina T Alamanou
is a great roundup of the AI Drug Discovery space. She does these every week, so if the intersection interests you, then she should be your go-to source.

Companion Guide to the Livestream

This guide expands the core ideas and structures them for deeper reflection. Watch the full stream for tone, nuance, and side-commentary.

Meta’s Strategic Realignment (00:05:10 – 00:15:30)

In case you missed it our deep dive on Meta is out now, and it breaks down the difference between Infra and Apps companies are different

Partnership with Midjourney
- Meta will integrate Midjourney for image and video generation inside its ecosystem.
- Rationale: Midjourney consistently ranks highest for aesthetics across the commercial image-gen space, though it lags in compliance/precision. For advertisers, aesthetics = conversion uplift.
- Implication: Meta’s prior quarterly outperformance already tied to generative ad tools; Midjourney integration compounds that edge. This could make ad creatives more compelling while further locking advertisers into Meta’s pipeline.
$10B Google Cloud Deal
- Meta offloading massive infra + inference workloads to GCP.
- Signal: Meta ceding the infrastructure race entirely. Choosing to concentrate on the application layer (social platforms, advertising, metaverse experiences) rather than sinking capex into infra battles with Google, Microsoft, Amazon.
- Cultural shift: Breaks the American tech tradition of “build everything in-house.” Mirrors the pragmatic Chinese playbook (Alibaba swapping to DeepSeek when results proved superior).

Grace Shao

has great analysis on the differences in AI Labs culture (I stole the above analysis directly from her), but I couldn’t find the article. Go give her a follow and read her work, she’s very very good.

Metaverse Alignment
- Both moves (Midjourney + GCP) enable scalable metaverse development.
- Constraint point: Infra demand from VR/AR worlds is outpacing Meta’s internal data center buildout. Outsourcing ensures supply.
- Bull case: If wearables (headsets, glasses) continue traction, Meta positions itself as the creative runtime for immersive spaces.

Google’s Government Offensive (00:16:30 – 00:24:10)

Gemini for Government – $0.47 per agency/year
- Includes NotebookLM, VIO (image gen), Gemini, FedRAMP compliance.
- Strategy: Loss-leader pricing to disintermediate service providers currently reselling LLM wrappers to agencies at $10K–$50K/month.
Strategic Outcomes
1. Market Reset: Forces out smaller resellers; creates a field where only hyperscalers can compete.
2. Lock-in: Once embedded, government workloads rarely churn; long-term pricing power follows.
3. Competitive Front: This leaves Google vs Microsoft as the only viable infra competitors. OpenAI/Anthropic lack profitability to fight in a price war.
Broader Implications
- Undercutting here isn’t about revenue today; it’s about distribution power.
- Establishes Google’s AI + productivity stack (Workspace, Gemini) as the default for public-sector digital transformation.

Apple’s Siri and Gemini Integration (00:34:30 – 00:46:00)

Siri’s Rule-Based Collapse
- Siri built on deterministic, rule-based frameworks. As LLMs rose, Siri’s architecture became brittle and outdated.
- Apple Intelligence rollout reinforced this weakness—widely criticized as shallow.
Considering Gemini Integration
- Talks of embedding Google’s Gemini to modernize Siri.
- Departure: Breaks with Apple’s aesthetic of vertical integration (owning every layer of the stack).
- Raises hardware implications: Gemini support would require TPU-class compute or specialized interconnects. Apple may be forced into deeper reliance on external chip ecosystems.
Risk/Reward
- Reward: Technical performance gains; consumers unlikely to care about outsourcing if the UX improves.
- Risk: Strategic dependence; Apple could slide further into financial engineering (buybacks/dividends) rather than innovating internally.
- Symbolic: An admission that Apple can no longer self-sustain AI competitiveness.

Reddit’s Emergence as AI Intel Hub (00:27:58 – 00:34:22)

Shift in Search Dynamics
- Reddit overtaking Google for AI-related queries.
- Why:
  - Google = slower, polished, curated knowledge.
  - Reddit = fast, raw, user-generated signals. AI moves too quickly for formal publication pipelines.
- Result: Early impressions, benchmarks, and leaks surface on Reddit before mainstream outlets.
Challenge for Reddit
- Value captured off-platform: startups and market analysts mine Reddit for signals (Ikidis included).
- Reddit itself struggles to monetize that flow; community anti-commercial sentiment resists API sales, bans, or licensing attempts.
- Structural irony: Reddit drives value creation elsewhere while under-monetizing its own informational dominance.

OpenAI, Anthropic & Strategic Diversification (00:24:15 – 00:33:00; 00:46:06 – 00:51:15)

OpenAI’s $30B Oracle Data Center Deal
- Funded via investor capital, not profits (OpenAI remains loss-making).
- Objective: build infra ownership, reduce dependency on hyperscalers, and expand investor narratives beyond “just models.”
- Coupled with chip-design hires from Google → signals push toward full-stack control.
Anthropic’s Wedge Play
- Investing in AI safety thought leadership and context-engineering research.
- Purpose: establish differentiation beyond models; build defensible narratives around safety and reliability.
Common Thread:
- Both firms shifting from single-pillar (model providers) to multi-pillar stories: infra, apps, chips, safety.
- Function: derisk future valuations by giving investors multiple reasons to fund continued expansion.

Data Center Debate: Oversupply vs Necessity (00:57:38 – 01:08:21)

U.S. View
- Current oversupply acknowledged; but infra buildouts justified because:
  - Infra supply often precedes demand (classic pattern in roads, grids, telecom).
  - Data centers serve as strategic signaling (“biggest data center ever” = investor confidence).
  - Assets can be repurposed for broader cloud workloads beyond AI.
- Problem: Externalities borne by local environments (fossil fuels, water depletion in desert states like Arizona).
Global South View
- ROI calculus misaligned: developing countries lack reliable grids/water, making DC buildouts wasteful.
- Smarter investments: education, renewable energy, open-source ecosystems.
- Parallel: Like startups outsourcing infra rather than building in-house models—developing economies should buy where possible, not replicate prematurely.
Conclusion:
- In rich economies → rational business case despite inefficiencies.
- In poorer economies → premature vanity project.

For those interested in the China side of the data center/cloud debate,

Grace Shao

(recommended twice in one roundup, that’s just how good she is) has a sharp write-up on how the AI boom could accelerate China’s cloud industry. It complements our discussion by showing how infra overbuild in the U.S. contrasts with infra-as-opportunity in China. Check it out here: “China’s AI Boom Could Be the Catalyst for Its Cloud Industry”

Research Spotlight – OpenCRISPR (01:08:28 – 01:11:48)

Project Overview
- Open-source initiative using LLMs to generate and edit enzymes with targeted properties.
- Potential: dramatically lower costs of gene-editing experiments by front-loading design with generative models.
Challenges
- AI strong at producing valid proteins/enzymes in silico.
- Weak at optimizing for utility—functional relevance inside biological systems is harder.
Takeaway:
- Promising avenue for scaling biomedical R&D throughput, but still early-stage.

Macro Themes Across the Session

Infra vs Apps Divergence
- Meta doubling down on apps, outsourcing infra.
- Google doubling down on infra, commoditizing competition.
- Apple caught between—outsourcing apps (Gemini) but lacking infra story.
Consolidation over Self-Sufficiency
- Tech giants increasingly partnering rather than vertically owning every layer.
- Marked shift from prior decade’s ethos of complete stack control.
ROI Misalignments
- Many infra investments (data centers, LLMs) are good for corporate signaling but questionable for societal value.
- Developing economies must resist vanity plays that mimic U.S. strategies without capacity.

Watch the full livestream for unfiltered commentary, side debates, and live audience questions that pushed these points even further.

Subscribe to support AI Made Simple and help us deliver more quality information to you-

Flexible pricing available—pay what matches your budget here.

Thank you for being here, and I hope you have a wonderful day.

Dev <3

If you liked this article and wish to share it, please refer to the following guidelines.

That is it for this piece. I appreciate your time. As always, if you’re interested in working with me or checking out my other work, my links will be at the end of this email/post. And if you found value in this write-up, I would appreciate you sharing it with more people. It is word-of-mouth referrals like yours that help me grow. The best way to share testimonials is to share articles and tag me in your post so I can see/share it.