Weekend Food For Thought WFFT
On today's menu: Kimi K3, Dinosaurs Becomes a Hot New Asset Class, Japan Progress Recovering Lithium, The Busy Person's Guide to Thinking, and more...
Hello from London,
I hope you had an interesting and productive week.
Claude, when prompted, shared the following on the systems thinking lessons one can take from sand dunes: “Sand dunes: self-organization without a blueprint - A dune has no architect. It emerges from the interaction of wind, particle size, and surface friction. Yet dunes develop consistent, recognizable forms — barchan, transverse, star — each one the signature of a particular wind regime. The form encodes the forces that made it. This is one of the deepest systems insights: structure is frozen process. When you look at a dune, you’re reading a history of wind. When you look at an organization’s hierarchy, or a city’s street layout, or a scar on a landscape, you’re reading the accumulated pressure of the forces that shaped it. The structure and the process are not separate things. Dunes also migrate. A dune moves downwind — not because any grain travels far, but because grains are continuously stripped from the windward face and deposited on the lee. The dune maintains its identity while constantly replacing its material. This is exactly what living systems do. You are not the same atoms you were seven years ago. The pattern persists; the substrate flows through it. The avalanche threshold - Sand piles build to a critical angle — the angle of repose — and then avalanche. Crucially, you cannot predict which grain will trigger the slide. The system accumulates stress invisibly, and release is triggered by something that looks arbitrary. The grain that starts the avalanche is not special; the system was ready. This is Per Bak’s self-organized criticality, and it appears everywhere: forest fires, earthquakes, financial crashes, extinctions. Systems driven by gradual inputs don’t fail gradually — they build to criticality and release catastrophically. The policy implication is uncomfortable: the longer you suppress small releases (small fires, small corrections, small conflicts), the larger the eventual avalanche.”
Take Note…watch the shifting sands…
1 Getting Visual
2 If You Read One Thing Today - Make Sure it is This
3 Consequential Thinking about Consequential Matters
4 Big Ideas
5 Big thinking
6 Keep On Keeping On…Upwards
1 Getting Visual
The Big Picture: Global Oil Transit Pressure Points - Watch the Strait of Hormuz & Bab el-Mandeb…
The Big Picture: The US “King of Oil (& NatGas)” & 2nd order benefactor of the last two ongoing conflicts…
Spotlight: “The Greatest Cash Machines Ever Built Just Went Cash Negative - Bank of America calls it a generational transfer in free cash flow, and the chart earns the phrase: twelve month forward free cash flow for the hyperscalers (Amazon, Alphabet, Meta, Microsoft and Oracle) is projected to turn negative for the first time in the dataset, collapsing from a peak near two hundred sixty billion dollars in 2024, while the semiconductor complex of Nvidia, Micron, Broadcom and Applied Materials surges past four hundred billion dollars, roughly triple its level of two years ago. The mechanism is simple plumbing: the hyperscalers are projected to spend on the order of $1.8 trillion dollars on AI infrastructure across 2026 and 2027, and a large share of every capex dollar exits their cash flow statement and lands, almost immediately, in the suppliers’ one. The lazy read is that the money simply moved and the semis won; the harder truth is that risk moved with it. Those $430bln of supplier cash flow are not independent income, they are entirely a function of customers whose own free cash flow has just crossed zero and who, per BofA, are now funding the gap with roughly $170bln of debt issuance this year, against an average near $30bln over the prior 5 years. Every prior capex mania, from railroads to telecoms to the Cisco era, followed the same script: supplier windfalls at the top of the cycle proved to be peak earnings rather than new baselines, because customer spending discipline eventually returned. The fair caveat is that negative free cash flow can be a deliberate investment choice rather than distress, and it is rational if the returns arrive; the question worth carrying into every valuation on both sides of this chart is which one already prices the transfer as permanent.” - Exponential View
Spotlight: The long view indicates the certainty of boom & bust cycle nature of the Memory Industry…watch the shifting sands…
Spotlight: The Chinese Models - Kimi can code. Kimi-K3, released last week, overtakes Fable 5 and GPT-5.6 Sol to take first place on the frontend code arena benchmark. And DeepSeek nears $500 million in annualized revenue with gross margins as high as 70-80% on its V4 model.
Spotlight: US VC and Startup Trends: Circular financing or smart allocations? “Corporate venture capital has become one of the most powerful forces in US AI startup funding, and it is increasingly concentrated. A new PitchBook analyst note, Fewer Deals, Bigger Bets, finds that corporate investors accounted for a record 87.9% of US AI VC deal value in 2026 so far, even as they took part in a shrinking share of deals by count, at 18.2%, the lowest in our dataset. Within that total, a small set of large-ticket investors is driving a disproportionate share of the value. The divergence, a rising share of value across a shrinking share of deals, points to a market where a handful of players write ever-larger checks into the most valuable AI companies. AI now makes up 93.6% of all corporate VC deal value, a decade high, as multibillion-dollar rounds for the likes of OpenAI and Anthropic pull corporate money toward the frontier. The note maps these investors into distinct groups. Cloud hyperscalers such as Amazon, Microsoft and Google more often invest to secure model access and lock in cloud commitments, while enterprise and infrastructure names such as Salesforce, Cisco, Qualcomm and Intel tend to invest to embed AI in their own products. The frontier labs, for their part, increasingly use their own capital as a go-to-market strategy. Standing apart is Nvidia, the single largest corporate investor by deal value, with a hand in $189.1 billion of US AI deals in 2026 YTD. It backs competing labs, including OpenAI, Anthropic, xAI and Mistral, with one aim: to keep the whole ecosystem dependent on GPU compute, regardless of which company or model ultimately prevails. What comes next will be shaped by a few open questions: the durability of the mega-round dynamic as OpenAI and Anthropic move toward public markets, the path of a shrinking CVC population, and the conversion of today’s strategic bets into durable commercial relationships.” - Pitchbook
Spotlight: Europe’s humanoid robotics ecosystem is much bigger than most people realize - “When people talk about humanoid robots, they often focus on the robot manufacturers. But the real strength of Europe lies beneath the surface—in the technologies that make these robots possible: Precision actuators and gearboxes. Dexterous hands and tactile sensing. Vision and perception systems. Power electronics and semiconductors. Functional safety and cybersecurity. Robotics software and AI. Advanced manufacturing and automation. This ecosystem builds on decades of automotive and industrial engineering excellence. Companies across Europe are already supplying many of the critical technologies that will define the next generation of intelligent, embodied AI systems.For me, the future of humanoid robotics is about building a resilient, scalable value chain that enables global innovation.The future of humanoid robotics will be built by ecosystems, not individual companies.” - Dirk Geiger
Spotlight - Real Assets: The collectibles - “Dinosaurs have become an asset class for billionaires. And the market is about as orderly and genteel as a starving T. rex - Dinosaur skeletons had recently become a hot new asset class, coveted by the ultrarich, not unlike sports franchises or Fabergé eggs. They offered a mix of glamour and primal appeal to a certain type of high-net-worth buyer—someone who might find the conventional art market fussy or convoluted but couldn’t deny the allure of a T. rex. “It could gobble you up in one bite. It’s just an amazing creature,” says billionaire Dan O’Dowd, who bought his T. rex, Samson, for a bargain $600,000 in 2009. “It’s the best trophy kind of thing you could own.” Two years before the Shen auction, Christie’s had sold a T. rex named Stan, said to be one of the most complete and best-preserved specimens ever found, for $31.8 million, trouncing the top-end presale estimate of $8 million. And two years after the Shen auction, financier Ken Griffin spent $44.6 million on a stegosaurus (“a vegetarian,” O’Dowd notes). Leonardo DiCaprio, Nicolas Cage and other celebrities have gotten in on the dino game too.”
Go shopping here: https://www.christies.com/en/stories/the-life-of-stan-a-t-rex-excavated-in-1992-f832007c6d9643e4ac214d5bee7ee5f7?utm_
2 If You Read One Thing Today - Make Sure it is This
Andrew McDermott guest writes for the Japan Optimist and uncovers how cooperating Salaryman CEOs quietly create more resilient value than limelight greedy Private Equity grand-man financiers - it’s worth a read in full - do it here:
Some Takeaways
“Congratulations to the investors and speculators in Kioxia. Your profits are your reward. The rest of us must decide what lessons to draw and how to apply them.
Kioxia’s market cap recently exceeded Toyota’s to rank as “Japan’s Number 1.” Toyota, the archetype of “old Japan Mochiai,“ has earned cumulative EBIT of ¥29T since 2018, earning more each year than the cumulative ¥1.2T EBIT earned by Kioxia over the same period. Has the torch been passed to a new generation?
Has Bain, the architect of the Kioxia deal, found the key to Japan’s future success: financially “disciplined” pure-plays freed from the tyranny of crossholdings, ably led by financial engineers trained in US business schools? Or is there more to the story: a series of quiet capital transactions between Japanese corporations that have refined Japan’s competitive edge, saving Kioxia in the process?
Is the future Bain’s “Asia Fund VI” or a third iteration of Mochiai: a series of intra-corporate transactions that make financial and strategic sense without sacrificing national security or excessively rewarding financial engineers? Let’s dig in.”
We are the heroes
Recently, we have heard a great deal from Bain about its Kioxia success and the implications for Japan and the world at large. The heroes of this story are clear: a team of financial engineers led by Mr. Sugimoto in Japan and Mr. Gross in Boston. They attribute success to “Bain Capital’s global technology investing footprint” that “was vital to understanding the business’s value proposition to customers and the nature and pace of its technology efforts.” Its weapons were not fear, surprise and ruthless efficiency, but “deep market knowledge, extensive local networks and [a] proven ability to drive operational improvement.”
Conveniently for Bain, it completed its interviews and its exit a few days before the announcement of a loss of an intellectual property case in the US, a case that began and ended under Bain’s leadership and whose announcement wiped tens of billions of dollars off Kioxia’s market cap (but not Bain’s profits)[1]. While Bain has been coy about the size of these profits., market observers speculate that the $2bln it staked in 2017 generated proceeds of $30-35bln, making it what the FT deems one of “history’s most lucrative” PE deals. This translates into ~$6bln of performance fees for Bain at an industry standard 20% of profits (exclusive of the 2% annual fee charged to clients and the ¥1bln/year charged to Kioxia itself to help “keep the lights on” at Bain). This heroic story comes at a convenient time for an industry whose press outside of Japan has been unflattering. Bain’s story has been well-told, and critics of the transaction (including yours truly) have been fed a steady diet of crow.”
What are we, chopped liver?
But could there be more to the story than the heroics of thirty or so MBAs and Stacey Smith, a former Intel CFO who seems to be the only person Bain placed in a management role?
We have heard nothing from the disparate group of METI bureaucrats, low-profile Japanese financiers and corporate engineer/executives who have, collectively, provided far more capital and taken far more Kioxia risk than Bain and its activist collaborators since the 2017 Toshiba recapitalization.
This essay attempts to give this group a voice. It seems fair that this voice should also receive the gift of hindsight that has characterized the triumphal Bain interviews that have conveniently coincided with Bain’s complete exit from its Kioxia position this past month.
While Bain is “all-out” of Kioxia, this team, like Warren Buffett, is committed to its Japanese investments for the long haul.”
Rick: Can you give some “Mochiai 3.0” examples other than Toshiba Memory?
Ray: Just off the top of my head: the recent Berkshire/Tokio deal, the ongoing Rohm/MEI transaction, the Sumitomo/MP Materials/DOD deal, and, most relevant to our discussion, the Rapidus and JSAMsemiconductor investments that cover the high-end and low-end of the semi chain. And, of course, deals that don’t happen fall into the same category in a different way. Examples include the MBK/Makino Milling deal, and the various PRC-linked efforts to buy Toshiba assets. The “good” deals are part of the ongoing optimization of Japanese companies that has driven the explosion of profitability that Jesper’s chart below illustrates. Notably, and despite what you might read, this has happened with very little influence from private equity, or, for the most part, “activist investors.” It has been an organic reaction to opportunities, risks and technology changes encountered in the real world, not to asset allocation models. When done properly, it’s remarkably efficient. That’s why investors like the GPIF can manage so much money so well with low fees and a small staff: companies do the work for them, a fact that Warren Buffett appreciates.”
“But I should hasten to add that there’s nothing uniquely Japanese about Mochiai 3.0. Examples include Microsoft’s investment in Apple, Apple’s own investment in Kioxia, Amazon (and Softbank’s) investments in OpenAI, and a series of investments made by Intel in its early years. In fact, the best explanation of Mochiai 3.0 was actually given by my hero Les Vadasz, the fourth employee of Intel. He led the design team that developed Intel’s first CPU (ironically, with the help of Japanese engineers). He also founded Intel Capital. His philosophy captures the essence of Mochiai 3.0: “you judge your returnon how you met your strategic goals, and count the money later,” but “the legal, financial and strategic consistency was always there.”
Rick: What’s the opposite of Mochiai 3.0?
Ray: The idea that asset allocation by financial engineers can create value at the portfolio company or national level without managers actually learning how to do the thing being managed.
A good overview can be found here. It is the replacement of men like Bill Knudson and Fred Burg with finance specialists who, even when they are right, create high frictional costs. Investment bankers are valuable, and I’m sure the bankers and consultants who work at the various investment firms trying to reshape Japan are very good at their jobs.
But, as America and the UK have shown us in recent years, a background in investment banking or consulting does little to prepare anyone for the exigences of running an actual business.
In fact, too much financial expertise can be dangerous because, as Sir George Buckley wrote, “In science and engineering-based companies, engineers know the art of the product possible. Finance majors don’t.”
Sir George personifies the spirit of Mochiai 3.0. He’s an engineer with multiple patents to his name, experience as CEO of 3M and chair of Smiths, and an outspoken critic of the “GE approach” (which is essentially the PE approach) before GE’s failings became obvious. That’s why Nakanishisan asked him to be the first foreign board member of Hitachi, which is of course the best example of Mochiai 3.0.”
“Rick: What’s your bottom line?
Ray: You run your own numbers. This is just an estimate because Toshiba doesn’t disclose its balance sheet. We do know that it took on about ¥1.2T of debt in 2023 and that it has sold its Kioxia stake down from 40% pre-IPO to 16% today. However, if you take market estimates for sale proceeds plus ~¥6.8T of Kioxia at market prices in early July and you apply a 15x EBIT trailing multiple to the rest of Toshiba (low relative to comparables) and you assume that cash flow over the past two years has paid down ¥200bln of debt, you can get to an equity value of ¥12-13T pretty easily.
Rick: So you’re suggesting that a ¥1T investment by “Mochiai 3.0” made three years ago could be worth 12-13x its original cost? And are you implying that, if Bain’s equity in Kioxia had been treated the way Toshiba was in 2017 rather than being repeatedly bailed out by taxpayers, banks and other equity players, then the Bain equity stake would have been considerably diluted if not wiped out entirely?
Ray: You’re saying this, not me. All I can say is that this transaction, like all other Mochiai 3.0 transactions, had strategic as well as financial goals. Yeah, we have made a lot of money on this so far, but, as you’ve pointed out, we are still invested in various ways.
There’s no exit for us here. As far as Bain’s (and Goldman’s) profits go, it was essential that they make a lot of money on this deal for reasons that have nothing to do with Kioxia per se.
Rick: Maybe this is a good time to back up and talk about the ends that the means of Mochiai 3.0 are supposed to serve.
Ray: That’s simple: national survival. Our prime directive for the last thousand years has been to preserve independence from China and Russia.
We faced a delicate situation in the mid 2010s: US elites, South Korea and the Philippines were all drifting into China’s orbit even as we watched the degradation of the Seventh Fleet.
We needed to do several things with the Toshiba transactions: win a few US elites to our side without compromising our manufacturing base, begin the hard work of reconciliation with South Korea and maximize the impact of our human capital on the areas that we (and China) knew would matter most in a decade.
These included energy abundance, semiconductor process technology and AI-enabled manufacturing for both civilian and military purposes.
The means have changed over the years, but the end remains the same: preserve our strategic autonomy, which can only be done by retaining the ability to make things that matter ourselves or with close allies.”
“Ray: I asked Mr. Vadasz what he thought about balance sheet strength in relation to semiconductor and other manufacturing companies. Here’s what he said:
In the old days the philosophy was driven by Gordon Moore, who really believed that people ought to be able to afford what they do. Anyway, we strongly believed that we should have a strong cash position. It was almost like a security blanket. Yet, we were a pretty egalitarian company. Most things we did were with no frills. We did not have assigned parking spots for executives. No separate eating places for executives, cubicles for all, including the CEO. No private jets.
Stock buy-backs never were big enough to jeopardize what we felt was an essential minimum. Frankly, I do not remember what we called a minimum at that time. Remember, that a wafer fab facility costs billions and leading edge tech companies need new facilities all the time.
One thing for sure: we were never, and I mean NEVER were guided by what the market felt was the right amount of cash to hold. It had to be comfortable from an operating point of view. All of us in top management roles would have defended this position, if needed.
A funny illustration of how extreme this went: in the early days we were proud of building and owning our own buildings. It took a lot of effort by one of the new CFOs to convince Gordon to sell the building, get the cash, and lease back the building. It was just not in his DNA. Anyway, we finally did it.
I do not know how to operate a company with the kind of leverage that some do. At times it feels like they are building a house of cards. If Japanese companies operate with more cash on hand, more power to them. It just feels intuitively obvious that they are more shock proof than the highly leveraged ones.
The old school wasn’t that bad… we were never limited by cash, just by our ability to create growth.
Rick: Given the above, how would you assess Bain’s stewardship?
Ray: I wish I could be more enthusiastic, and I hope that I’m wrong. Perhaps Bain’s employees brought patents or operating expertise similar to what Pat Gelsinger brought Intel–real technical skills that will pay off in the years to come.
What’s certain is that they have left Kioxia shockingly undercapitalized relative to peers.”
“These firms, all of whom are led by experienced operators, understand the capital intensity, technical pace and cyclicality of this business. While Bain has been using this hot market to sell shares on its own behalf, these peers have been preparing for the next cycle by strengthening their balance sheets. The amount SK and CXMT are raising this month for operations approximates the amount Bain has taken for itself in equity sales.
It would be one thing if Bain’s exit had come after Kioxia had surpassed its competitors in the marketplace and had a balance sheet that was at least in line with, if not stronger than, its peers. The opposite is the case: Kioxia has lost share in its core business in each of the last four quarters, its technical edge in NAND is shrinking, and it has completely ceded the HBM market to peers, including an upstart Chinese competitor who has done what (supposedly) was impossible: start a NAND and HBM business from scratch.
Curiously, in its capacity as the controlling shareholder, Bain (somewhat reluctantly) agreed to an IPO price of ¥1,455. At this price, Bain allowed the company to sell 21.5mm new shares while selling only a few of its own shares. Yet, as the price increased above ¥75,000, Bain was a seller of its own position without allowing the company to raise capital for itself alongside its competitors.”
“Rick: This seems like a good time to talk about the SK relationship. Can you discuss?
Ray: At the company level, it was very complicated. SK is a strategic rival. Viewed narrowly, including them made no sense because SK could (and did) use its position to influence management decisions. However, in the context of Japan/Korea relations, we’re cautiously optimistic. Our human capital is limited. We have major projects at the high end (Rapidus) and the low end (JSAM). SK’s balance sheet is where ours needs to be. Our hope is that five years from now, we find a way to deepen the SK partnership, perhaps swapping NAND capacity for high-end technology developed by Rapidus and/or in conjunction with TSMC. This is one way for us to confront the PRC onslaught. It fits into the broader picture of increased military, energy and diplomatic partnerships we are forming with Korea and could be a critical component of making Tokyo a regional financial center.”
“…we’ve taken steps to ensure that Mochiai 3.0 remains in control of critical assets while allowing PE firms to bid against each other for assets that don’t matter very much (much as America let Japanese firms bid over the odds for Pebble Beach and Rockefeller Center). This has predictably elicited calls of “backsliding,” but we think the problem will take care of itself as the returns come in from the deals we did not backstop, like Calsonic (twice bankrupt) and PHC Holdings (five CEOs, multiple scandals). When investors compare the risks and returns of these transactions to what “plain old Japanese public companies” offer at a fraction of the cost, we think there’s a good chance that a “new Tokyo” financial center model might emerge. The deals that work best tend to be those with in-place Japanese management, like the Hitachi spins to both public and private equity. A broader public market will absorb most of these companies directly via spin offs that don’t require a trip through the PE cleaners. This is what Sony did with Sony Financial and what Western Digital did with its own memory business (an inconvenient truth for Bain).”
“The “Tokyo model,” if adopted, will be one that recognizes and promotes the idea that the asset management industry is an important supporting actor in any economy, but cannot ever be the creator of value.
Keeping this balance is important. In the long-run, as both Palmer Luckey and Chris Miller (author of Chip Wars) argue, our marginal “incentive” dollars will be best spent not on investment fees but on educating more engineers ourselves and resurrecting the “defector visa program” to get some of China’s best players on our team. Imagine if we offered $6bln to the top 30 engineers at CXMT rather than to the top 30 partners at Bain?
If this sounds crazy, it’s not. This is the playbook the CCP is running right now, and it’s the same one that helped us win the Cold War. There are over 2 million Chinese living in Japan, including some of the most innovative tech founders. Let’s get more.”
“I was raised in an environment conducive to making things. Right now, there’s a fearful symmetry between the financial engineering talent flowing from the US to Japan and the actual engineering talent flowing in the other direction.
The US, while sending legions of investors intent on “fixing Japan,” is asking Japan not just to pay for rebuilding America, but to show America how. Supposedly “inefficient” Japanese companies are sending engineers to train Americans in everything from making steel to nuclear power plants to shipbuilding to machine tools to quantum computing to robots.
The decline in America’s ability to make things has been catastrophic, not just for America, but for us. Compounding this problem, there has been a generational collapse in the US/Japan education exchange even as US elite educational and financial institutions continue to spend energy courting China.
It’s very curious to see that China is also trying to attract our engineers while seeming to have more than enough bankers.”
“The lesson from history and industry is clear: leadership by engineers and builders produces lasting strategic strength; leadership by financiers alone optimises for short-term gains at the expense of long-term readiness. David Murrin
We’re throwing more and more of our resources, including the cream of our youth, into financial activities remote from the production of goods and services, into activities that generate high private rewards disproportionate to their social productivity. Nobel Prize-winning economist James Tobin quoted in: When the Machine Stopped: A Cautionary Tale from Industrial America“
“Now the only defense that any nation can have is the character and intelligence of its people.
The adequacy of that defense will depend upon the strength of the conviction that the nation is worth defending.
We have repeated to ourselves so much of late the slogan “America must be strong,” that we have forgotten what strength is.
We appear to believe that strength consists of masses of men and machines. I do not deny that they have their role. But surely the essential ingredients of strength are trained intelligence, love of country, the understanding of its ideals, and such devotion to those ideals that they become a part of the thought and life of every citizen.” Robert Hutchins, The Great Conversation.
3 Consequential Thinking about Consequential Matters
Kimi K3 dropped last week, and days later Xi Jinping gave his most substantive speech on AI yet at the World AI Conference in Shanghai. The full ChinaTalk squad convened for an emergency pod on what a Chinese Mythos-class model would mean, and whether Beijing has any mechanism to catch one before it ships open-weight. You can explore that conversation on consequential matters here in full:
The ChinaTalk team asks: “When a Chinese lab trains a model to match Claude Mythos — the system whose cyber capabilities pushed Washington into a de facto licensing regime — what will Beijing do?” They and Kimi maps out the different paths and seeks to answer that consequential question here:
https://china-mythos.vercel.app/?utm_
Some Takeaways
“It’s a big deal because now everyone in the US AI community is a little bit afraid. In the past months we were pulling farther and farther ahead from China in terms of model quality and model size. And first it was GLM 5.2, and now this week Kimi K3 just shattering that illusion. Kimi and China may be closer to a Mythos model than we think.”
“I think it’s interesting the extent to which other countries are being talked about in the context of the expansion of the AI ecosystem, and the cooperation initiatives announced with Belt and Road countries, the Shanghai Cooperation Organization, ASEAN. When I was in China touring some of the Chinese AI labs, I found out that a Chinese company is helping Kazakhstan build their sovereign AI initiative.”
“I want to stay on Lily’s point that this feels like a speech for the world. It reminded me a lot of the Xi Davos 2017 one, the we’re-open-for-business speech that opened with a Charles Dickens quote. Within weeks there was THAAD in South Korea and a campaign of economic coercion against that country, and in the subsequent years, tiffs with Japan and Australia and India on down the list.
In terms of reception, I was at a conference with some American AI investors, and all of them were like, wow, this is remarkable. I wish my president could put out this vision to the world. I would just encourage folks, if this is the only Xi speech they’ve ever read in their life, to maybe take it with a bit of a grain of salt — these promises of global comity and an outstretched arm often have other layers and subtext to them.”
“I think the lens that AI is just normal technology actually explains a whole lot when you think about this speech. As an American audience member, you sometimes feel like there’s this weird tension between wanting AI to be open and inclusive, but also safe and secure. We think those things are inherently contradictory. But the Chinese position on this is like, look, it’s just like the internet. You want it to be as diffuse as possible, but just have good cybersecurity. What’s the problem? Why can’t you do both? We’ll see how and if that changes in a post-Mythos world.”
“And we don’t really want to buy American chips even if they’re offering them, because if your timelines are longer, then China’s done without for a while with lots of other things in the past, for the goal of import substitution. So sorry, ByteDance, sorry, Alibaba, just deal with it. Make your own chips. We’ll get there.”
“I don’t know if the government necessarily cares so much about openness as an abstract value, but you have to remember that the people in the audience are all domestic Chinese AI researchers. And at some point there’s the question of, is the crackdown coming? The way that they came for Jack Ma, are we next? When Xi says openness is what we care about the most, maybe he’s just trying to say the government is on your side. Unclear how long that will last. He’s trying to hype up the domestic AI ecosystem and make them feel like they can innovate without being constrained too much.”
“I agree with Lily, I think he’s offering a compliment to open source and being like, yes, we’re back, for now.
I also want to hit the point about AI as normal technology. Xi ends with this quote, 明者因时而变，知者随事而制, which is officially translated as “a man of wisdom adapts to changes, a man of knowledge acts by circumstances.” I wanted to look at where it was from. It’s from this record of policy debates from the Western Han dynasty, around 81 BC. And I thought it was actually really interesting because this document is called 盐铁论, Discourses on Salt and Iron. A lot of it was about how the Western Han was governing the salt and iron industries, which by that point were largely privatized, and they were trying to figure out how much should we extract from this private enterprise. If we have any listeners who are experts in the Western Han, please correct me on this, but it’s a cool parallel with AI today, which in China is also largely privatized, and the state is trying to figure out how much to extract from it.
Jordan Schneider: And other empires brought it in house, right? There was a salt monopoly for hundreds of years at some point.”
Let it Rip...
“The Let It Rip scenario is just the status quo — the Chinese Mythos model will come out just as Kimi K3 did, just as GLM 5.2 did. It just drops. And maybe hopefully there’s a one-week delay saying we’ll release the weights later. But there doesn’t seem to be any system within the government for catching, this model’s different, this model actually has the capability to do some real damage. And then once it’s out, it’s out.
And maybe after that, this will be the wake-up call they need to scramble and have some policy to change how they review AI models. But an open-weight Kimi K4 or a GLM 6 can do some major damage out in the open.
Jordan Schneider: The question I have now is, what if it just doesn’t? What if it’s just another day in the life? We’ve had so much — can we call it hyperventilating? Can we call it alarm raising — from cybersecurity experts. There is a world in which they are wrong.
Let’s make the case. What was Kevin Xu saying? You need serious access to compute in data centers. Maybe at the cloud layer and the neoclouds, they just decide that they’re not going to do business with cyber hackers or Boko Haram or whoever else. Maybe things get patched fast enough that it’s just kind of a blip and we get to the other side, and instead of taking two or three years as some people are proposing, it takes a month. Then all of a sudden our models are great because there’s just a lot more money being put into these models to clean up systems than there is to attack systems. Other arguments folks have in favor of we just kind of move on?
Lily Ottinger: From the Chinese government perspective, they already live in a world of cyber exploit abundance. They’ve had a system for systematically acquiring zero-day vulnerabilities for a long time and have decided that the risks of using them outweigh the benefits in a lot of scenarios, but that’s because China’s a state actor with constraints placed on them by the multilateral system, the trade system. They don’t want the backlash.
And they feel like they have domestic insurgencies under control. So to them it’s not a threat domestically. I don’t think we are necessarily prepared for a world in which state actors — Claude Code made a lot of problems look software-shaped. I don’t think we’re prepared for a world in which a lot of problems look like they have solutions that are cyberattack-shaped. Unclear what steps are required to execute on any of those things. But the fact that the exploits exist, I feel like they’ll at least be sellable in a way that makes a lot of money for rogue actors in a way that we’re perhaps not prepared for. It’ll certainly be a nice way for the North Korean government to make money, but China’s not worried about them either.
You could be right that there’s no problem. I think the Chinese government for now certainly thinks it’s not a risk.
Aqib Zakaria: I think it could be such good PR for the Chinese government if it comes out without a major incident. Because people are upset over the safeguards on Fable 5 and all these lobotomized models, and now China’s the one giving it to us straight from the source. And then you have this scenario where the leading AI models are Chinese, and also the lower-cost ones are also Chinese, which could do major benefits for their adoption levels.
The Kevin Xu argument is that for Kimi, which is over two trillion parameters, you need like a dozen GPUs. You need a couple million on you, which is a lot, but certainly not insurmountable if you’re any organization or any evil rich guy, I suppose. It’s just a question of what actually gets done with it, and how much damage the CCP is willing to take. And if there is a major cyberattack with this Chinese Mythos model, how bad of PR is it for the government? And what can they even do after the fact to correct it?”
The Black Box
Jordan Schneider: All right, maybe we should jump from Let It Rip all the way to Black Box. Irene, what was your vision of the future?
Irene Zhang: My vision, which I think everyone ranked as the least possible, but which I will try to defend here, is a world where the Chinese state realizes that one of the leading labs in China develops a model that heavily surpasses previous models in cybersecurity capabilities and capabilities in general, and decides that rather than letting the company release it openly, they will get first dibs on it for government use and then only allow the company to release a lobotomized version, because it is too concerned about safety. And in that process partially nationalizes this lab and tries to coerce it into developing products that are more tailored towards government use and potentially offensive use.
The reason I think that’s a genuinely possible scenario — I know the biggest counterargument is that this is commercially bad for Chinese AI’s influence in the world, to have its top capability models not be available to any national audience to diffuse and compete with the US. But China’s advantage in some ways is that it has a lot of labs. Let’s say Kimi gets there first. Let’s say K4 is somehow extraordinary and the government takes that in house and forces Kimi to not release openly. Zhipu AI will get there someday, Qwen will get there, ByteDance — we don’t hear enough about ByteDance, but they will get there someday.
In a sense, the advantage that Beijing will get from a strategy like that maybe outweighs the short-term PR game and commercial game. My scenario makes sense in a world where the Chinese government is a lot more quote-unquote safety-pilled than people assume, and also sees genuinely useful things they can do with AI for national security and its strategic vision in the short term. Enough that it thinks making that sort of very explicit move to acquire top intelligence is really useful.
Jordan Schneider: Two points in your favor, Irene, from the Xi speech. Perhaps the most interesting paragraph was him saying we must attach great importance to the various endogenous and derivative risks brought about by AI and work to build systems of laws and regulations, technical monitoring, early risk warning, and emergency response. We must reinforce the safety baseline, prevent misuse and abuse, and ensure that AI always remains under human control. So clearly some safety emphasis there.
Plus, I gave the top three Chinese models and the top two American ones all three of our scenarios, and the Chinese models believed in the black box. Even though me, Lily, and Aqib all gave it a fifteen percent chance, DeepSeek was up to thirty percent, Kimi was up to forty-five percent. So the Chinese models think they’re going to get nationalized sooner rather than later.”
China’s Glasswing
Jordan Schneider: Lily, let’s go to you then. How does China’s Glasswing happen?
Lily Ottinger: Project Glasswing would happen in a world where the capabilities are flagged in advance by somebody. Someone from the Chinese government gives a speech in advance that says we need to move the testing capabilities up, we need to evaluate first, release later, we need to fold in relevant stakeholders from private industry and government ministries before releasing a model with huge step-changing capabilities. And then the labs take that as a signal that now it’s their responsibility to figure out whether their model is going to be Mythos level before they release it. Perhaps the fact that Kimi is waiting to release the open weights of K3 is an indication that the labs are thinking about this more.
I think this is pretty likely because there’s a playbook for it already, in the form of Project Glasswing. It enables the model to eventually still go out in a way that is safe for everybody and has the cyber capabilities lobotomized, but enables Chinese developers to still work with the best tools, which is something the Chinese government seems to really care about because they believe AI is a huge unlock as far as productivity is concerned.
So I think there are forces in the Chinese government that will work against having the models be totally black-boxed and government-only and never released.
I think it would cause a huge amount of infighting. The labs would not necessarily enjoy being the national champion of a Glasswing-style project, because it would mean handing over their proprietary technology to their competitors, all of which are building their own foundation models too. When Mythos was handed over to Microsoft, it wasn’t such a big deal, because Microsoft’s not a serious competitor in the AI race. But that’s not true in China, because everybody and their mother has their own foundation model, including delivery apps like Meituan. And they all hate each other and have longstanding rivalries and they all poach employees from each other. There’s so much drama.
Irene Zhang: I tried to think of examples previously of Chinese big tech cooperating with each other on stuff. And I really couldn’t think of any.
Lily Ottinger: And I think the government likes it better that way. They prefer that the companies don’t necessarily get along and do these big cooperation initiatives that stretch across conglomerates.”
4 Big Ideas
Techspot shares a quick look at Japans progress with methods for recovering lithium from old EV batteries - a key potential step for a circular economy model and to retain some level of self sustainability for a key input in the global economy - Read it full here:
https://www.techspot.com/news/112051-japan-finds-way-recover-90-lithium-old-ev.html?utm_
Some Takeaways
“THE BIG PICTURE: Countries that don’t mine their own lithium have two options when supply gets tight: pay whatever the market demands, or figure out how to reuse what they’ve already got. Japan is trying to lean more toward the latter, and one major breakthrough may help with that.
A facility in Fukui Prefecture has figured out how to extract a whopping 90% of the lithium sitting inside dead EV batteries. That’s around double what such operations previously achieved.
Behind the breakthrough is JX Metals Circular Solutions, a subsidiary of one of Japan’s largest non-ferrous metal companies. While it was announced back in April 2025, it really started grabbing headlines this month after some Japanese publications revealed the actual process at the company’s plant in Tsuruga.”
“The process starts with old batteries being separated and burned to strip away non-metal components. What’s left gets crushed into something called black mass. This is essentially a powder packed with recoverable metals. From there, a water-based chemical treatment called hydrometallurgy pulls the lithium out.
One clever distinction in this new process is that the recovered lithium hydroxide actually replaces a chemical traditionally used during refining. This cuts the carbon footprint by about 40% compared to older methods.”
“For Japan, this breakthrough is especially important since the country has so far imported virtually every bit of mineral it needs for its batteries. That includes not just lithium, but also cobalt and nickel. A lot of the refining has historically gone through China.
Japan has been trying to find a way around this. A new law taking effect this year will require manufacturers and importers to collect and recycle small portable batteries from the likes of phones, vapes, and power tools. The government wants recyclers to hit 70% lithium recovery by 2030, so that 90% number is already well ahead of schedule.
That said, Japan isn’t the only country pursuing lithium recovery. In the US, Redwood Materials – the recycling company founded by former Tesla CTO JB Straubel – says it’s already recovering 95% of lithium from the equivalent of about 250,000 EVs per year.
In Japan’s case, though, the biggest bottleneck right now isn’t the technology. It’s actually getting dead batteries to recyclers in the first place. Only about 14% of end-of-life lithium-ion batteries in the country currently make it through official collection channels. Many retired EVs actually end up getting exported, making those valuable metals inaccessible. Solving this problem is now more important than ever.”
5 Big thinking
The good people at FS Blog shares “The busy person’s guide to thinking” - it’s as per usual a good read & ponder session for you to explore - Go read it in full here:
Some Takeaways
“No skill is more valuable or harder to come by than the ability to critically think through problems.
Thinking better than others means you’ll have more free time and fewer problems. If you can’t think well, you’ll spend a lot of time fixing avoidable mistakes.
If you stop thinking when most people stop thinking, you’ll come to the same conclusions most people come to. Thinking is hard, so it’s natural that most people stop as soon as possible. We get the gist of something and move on to the next problem.”
“Thinking means concentrating on one thing long enough to develop an idea about it. Not learning other people’s ideas, or memorizing a body of information, however much those may sometimes be useful. Developing your own ideas. In short, thinking for yourself. You simply cannot do that in bursts of 20 seconds at a time, constantly interrupted by Facebook messages or Twitter tweets, or fiddling with your iPod, or watching something on YouTube.
I find for myself that my first thought is never my best thought. My first thought is always someone else’s; it’s always what I’ve already heard about the subject, always the conventional wisdom.
It’s only by concentrating, sticking to the question, being patient, letting all the parts of my mind come into play, that I arrive at an original idea.
By giving my brain a chance to make associations, draw connections, take me by surprise.
And often even that idea doesn’t turn out to be very good. I need time to think about it, too, to make mistakes and recognize them, to make false starts and correct them, to outlast my impulses, to defeat my desire to declare the job done and move on to the next thing.”
“So it is with any other form of thought. You do your best thinking by slowing down and concentrating.”
“The best way to improve your thinking ability is to spend time thinking.
Immersing ourselves in a subject and giving it our full attention offers insights that can’t be gained simply by skimming on the surface.
How can we force ourselves to spend more time thinking?
One way to engage with an idea is to write about it.”
“While we naturally understand that writing helps share ideas with others, we under-appreciate how it sharpens our own thinking. Writing isn’t just communication – it’s a tool for deeper reasoning. In a world of fragmented attention, writing forces us to slow down and think systematically.
Writing is the process by which we realize we don’t understand and the process by which we come to understand.
Writing requires compressing an idea. When done poorly, compression removes crucial information. When done well, compression keeps the essential and removes the rest. This combination of thinking and clarity is what makes writing such a powerful tool for thought.
“ If you’re thinking without writing, you only think you’re thinking.”
LESLIE LAMPORT
“Writing is thinking made visible. It exposes gaps in reasoning that casual thinking misses and creates a permanent record of insights that would otherwise fade. While sharing writing with others adds valuable perspective, the real power lies in how it forces us to wrestle with our own ideas until they become clear.
Start small. Take one important idea and explore it through writing. Your thinking will sharpen with each word.”
6 Keep On Keeping On…Upwards
Have a Great weekend when You get there…
Sune