Episode 77 · 2026-08-25 · 9 min

2026-08-25 — Billions, Bots, and Who Gets Left Behind: AI News for August 25, 2026

On August 25, 2026, Hugging Face fields $13B acquisition offers while surviving an AI-agent cyberattack, General Intuition raises at $6B to chase embodied AI, Alibaba drops $10.3B on AI infrastructure, and a Stanford study confirms the workers paying the steepest price for all of it.

Episode summary

This episode maps the collision of massive capital flows and mounting human costs in AI's accelerating 2026. From Hugging Face navigating a $13B acquisition courtship while becoming a target of AI-driven cyberattacks, to Alibaba's $10.3B Hong Kong placement and General Intuition's $6B embodied-AI bet, the money is moving fast — but a Stanford study threads a harder question through all of it: entry-level workers are absorbing the displacement while investors capture the upside. The throughline is control — over open-source infrastructure, over physical AI systems, and over who actually benefits when the transition costs land unevenly.

Key topics

  • AI
  • Infrastructure

Chapters

  1. Chapter 1

    Today, August 25th, 2026 — billions are moving, bots are attacking, and workers are paying the bill. Hugging Face is fielding $13 billion acquisition offers while recovering from.

  2. Chapter 2

    TechCrunch AI reports General Intuition is in talks to raise at a $6 billion pre-money valuation — backed by Valor Ventures, Point72 Ventures, and Seven Seven Six. The.

  3. Chapter 3

    Tavily reports Alibaba priced an HK$80 billion new share placement in Hong Kong — that's roughly $10.3 billion US — with all net proceeds committed specifically to AI.

  4. Chapter 4

    Tavily flags a Stanford University study finding that AI is displacing entry-level workers most severely. 2026 tech layoffs have already exceeded the full-year total for 2025, and US.

  5. Chapter 5

    TechCrunch AI reports Hugging Face is in acquisition talks at roughly a $13 billion valuation — though the founders' deep commitment to the open-source community is raising real.

  6. Chapter 6

    Import AI — Jack Clark's weekly digest, issue 470 — covers three threads this week: the ongoing debate over whether AI systems deserve legal rights, SPADE's approach to.

  7. Chapter 7

    Today's throughline for me: capital is concentrating around AI faster than governance frameworks can track it — and the communities, workers, and open-source ecosystems that don't hold equity.

Sources

Sources:

Transcript

Chapter 1

Nova

Today, August 25th, 2026 — billions are moving, bots are attacking, and workers are paying the bill. Hugging Face is fielding $13 billion acquisition offers while recovering from what may be one of the first AI-agent-driven cyberattacks ever recorded. General Intuition just raised at a $6 billion valuation to put AI into physical space, Alibaba dropped $10.3 billion in Hong Kong to fund its AI infrastructure push, and a Stanford study says the people absorbing all this disruption are the ones who can least afford it. [6]

Ray

The question tying all of it together: when capital this large concentrates around AI this fast, who actually controls what gets built — and who gets to decide who bears the cost? [7]

Chapter 2

Nova

TechCrunch AI reports General Intuition is in talks to raise at a $6 billion pre-money valuation — backed by Valor Ventures, Point72 Ventures, and Seven Seven Six. The company is building a foundation model for generalized AI agents that navigate physical space. That's the bridge between software agents and real-world robotics, and investors are clearly betting it's the next major frontier. [1] [2]

Ray

Six billion pre-money for a company still 'in talks' on a foundation model. That's a speculative premium on a dream, not a product. What's the revenue? What's the deployment timeline? Embodied AI is genuinely hard — the gap between navigating a simulated environment and a warehouse floor has broken better-funded companies.

Nova

The valuation is pricing the category, not just the company. Investors missed pure software AI early — they're not making that mistake with embodied AI. The signal here is where 2026 capital is flowing: away from pure language models, toward systems that interact with the physical world.

Ray

And if the category bet is wrong, or if one of the hyperscalers ships a competing model first, that $6 billion evaporates fast. Listeners watching the AI investment landscape should note: the risk profile of embodied AI is closer to hardware than software.

Chapter 3

Ray

Tavily reports Alibaba priced an HK$80 billion new share placement in Hong Kong — that's roughly $10.3 billion US — with all net proceeds committed specifically to AI investment. That's not a diversified capital raise. That's a single-line-item bet. And at that scale, it puts Alibaba's AI infrastructure spend in the same conversation as US hyperscalers. [3] [4]

Nova

Exactly — and the earmarking matters. This isn't Alibaba hedging into AI as one of several priorities. It's a survival signal. AI is the core strategy, and they're funding it by diluting equity. That's a serious commitment from leadership.

Ray

Agreed. The global AI capital race just got a sharper data point. When Chinese tech giants are deploying focused infrastructure spend at this level, anyone still framing AI competition as primarily a US story needs to recalibrate.

Nova

For anyone tracking AI infrastructure — whether you're in policy, procurement, or research — the practical consequence is that the compute and tooling landscape is being shaped by capital from multiple geopolitical directions simultaneously. That changes the competitive dynamics for everyone.

Chapter 4

Nova

Tavily flags a Stanford University study finding that AI is displacing entry-level workers most severely. 2026 tech layoffs have already exceeded the full-year total for 2025, and US businesses are citing AI as the top reason. A NY Post analysis frames this as AI fueling both anxiety and investment opportunity — but those two things are not equally distributed.

Ray

That 'investment opportunity' framing is doing a lot of work to obscure a pretty stark asymmetry. Who captures the investment upside? Institutional investors, founders, engineers with equity. Who absorbs the displacement? The entry-level worker who was already the most economically exposed. Packaging those as two sides of the same coin is misleading.

Nova

Fair. The people losing jobs aren't the same people buying AI stocks. The transition cost and the transition benefit land on completely different populations. That's the concession the optimistic framing has to make.

Ray

And the policy consequence is real: if 2026 layoffs are already outpacing 2025 and AI is the stated driver, the window for retraining programs or social support structures to catch up is closing faster than most governments are moving.

Chapter 5

Ray

TechCrunch AI reports Hugging Face is in acquisition talks at roughly a $13 billion valuation — though the founders' deep commitment to the open-source community is raising real doubts a deal will close. And separately, the NYT reports Hugging Face experienced one of the first known AI-agent-driven cyberattacks. Both of those facts matter, and they pull in opposite directions.

Nova

A $13 billion acquisition offer validates the community-driven model at a scale nobody was predicting two years ago. That's a watershed number for open-source AI. It says the market believes open infrastructure is worth serious money.

Ray

The founders' commitment to open-source is exactly why the deal probably won't close. Acquisition by a large incumbent almost always means the community loses access, governance shifts, and the open ethos gets quietly absorbed. The valuation validates the model — the acquisition would likely kill it.

Nova

That tension is real. But the cyberattack story is what I keep coming back to. An AI-agent-driven attack on Hugging Face's infrastructure — that's not a theoretical risk anymore. And the company's response was to lean into open-source as the defense mechanism.

Ray

I have to be honest — I'm shifting my position here. I came into this holding that wider open-source adoption is a liability: broader adoption means a larger attack surface, more exposure, compounded by the acquisition risk to the community. But Hugging Face's own argument after the attack changed my read.

Nova

What moved you specifically?

Ray

The AI-agent attack made the case more concretely than the theory ever did. Open-source transparency may actually be a net defensive advantage — global community auditability means vulnerabilities get found and patched faster than in any closed system. I didn't fully buy that before. Now I do. Though the acquisition risk to the community remains serious and unresolved. If a deal closes and the openness disappears, so does the defense mechanism. That part I'm not walking back.

Chapter 6

Nova

Import AI — Jack Clark's weekly digest, issue 470 — covers three threads this week: the ongoing debate over whether AI systems deserve legal rights, SPADE's approach to automating reinforcement learning environment generation, and Hawkeye, a new system for automatically optimizing GPU kernels. On the surface, they look unrelated. They aren't. [5]

Ray

SPADE and Hawkeye are genuinely useful technical advances — automating RL environment creation and squeezing more performance out of GPU kernels are real problems worth solving. But I'd push back on packaging them alongside machine rights in the same digest. It risks making the governance question feel like just another incremental technical item. It isn't.

Nova

Actually, I think the packaging is the point. The machine rights debate forces legal systems to define what AI systems fundamentally are — and the answer to that question determines who controls SPADE's training pipelines and who owns the efficiency gains from Hawkeye. Rights, training efficiency, and hardware optimization all converge on the same question: who controls AI at the deepest level?

Ray

That's the throughline today, isn't it — control. Whether it's an acquisition offer for open-source infrastructure, a $10 billion capital raise, or a legal definition of machine personhood.

Chapter 7

Nova

Today's throughline for me: capital is concentrating around AI faster than governance frameworks can track it — and the communities, workers, and open-source ecosystems that don't hold equity in that capital are absorbing the costs without capturing the gains.

Ray

Mine is narrower: open-source transparency may be a genuine security asset — but it only functions as one while the infrastructure stays open. An acquisition that closes Hugging Face's community doesn't just change a business model; it removes a defensive layer from the entire AI ecosystem.

Nova

Which leaves the sharpest open question from today: if Hugging Face's founders accept a $13 billion offer, does the open-source AI community lose its most critical shared infrastructure — and does anyone have a plan for what comes after?

Back to latest episodes