GUY: Good evening, and welcome to Morning Signal for Thursday, August thirteenth, twenty twenty-six. We have a late edition today, but it is a packed one: currency intervention, a labor market that looks very different depending on whether you already have a job, and a big shift in how investors should think about artificial intelligence.

AVA: And the connecting idea is capacity. Can Japan turn a dramatic currency intervention into durable policy? Can the United States build data centers without enough electricians? Can a robotaxi turn impressive autonomy into acceptable returns? We are going markets and macro first, then technology, geopolitics, the cross-currents, and a very specific watch list.

GUY: Let us start with Goldman Sachs Exchanges. Karen Fishman and Praneet Shah said Japan's Ministry of Finance may have sold as much as eighty-five billion dollars on July thirtieth and thirty-first to buy yen, with another possible twenty billion on August third. Average daily dollar-yen volume was roughly thirty billion, while EBS spot activity reached about ninety billion and eighty billion on those first two days. So this was not a gentle signal. It was enormous relative to normal market flow.

AVA: Goldman Sachs Exchanges also said dollar-yen fell about three percent during the Japanese action and another two percent after evidence of United States coordination, although the net move had retraced to about three percent by the recording. The American contribution was likely far smaller, historically around one to two billion dollars, but its signaling value was large. That distinction matters: Japan supplied the balance sheet, while the United States supplied credibility and reassurance about market functioning.

GUY: Right. Goldman Sachs Exchanges described the Federal Reserve's FIMA repo facility as part of the mechanism. Japan can raise dollars against Treasuries rather than dump those Treasuries abruptly into the secondary market. That helps explain why United States involvement matters even if its intervention dollars are modest. It lowers the risk that currency defense becomes disorderly Treasury selling.

AVA: But Goldman Sachs Exchanges was equally clear about the limit. Intervention forced leveraged carry and CTA positions to reduce exposure, and CFTC data showed the fourth-largest absolute reduction in yen positioning in twenty years. Yet flow can buy time; it cannot permanently override fiscal and monetary policy. The market was pricing a sixty-five percent probability of a twenty-five-basis-point Bank of Japan hike in September, and roughly forty basis points of tightening through year-end.

GUY: Which creates a clean test from Goldman Sachs Exchanges. If the Bank of Japan fails to validate intervention with a hike, renewed yen weakness becomes more likely. If it hikes and Japanese investors begin repatriating assets, appreciation has a more durable mechanism. The levels mentioned were a two-hundred-day moving average near one hundred fifty-eight, dollar-yen around one hundred sixty, and euro-yen near one hundred eighty-seven point five zero.

AVA: Goldman Sachs Exchanges also reported a genuine split rather than a consensus. Some clients were targeting dollar-yen near one hundred fifty and about five percent more yen strength against the euro. Others saw the intervention move as another chance to buy the dip. Karen Fishman expected modest dollar strength versus low-yielding currencies and weakness versus higher-yielding emerging-market currencies in a low-volatility carry environment. Praneet Shah instead expected a steadier broad-dollar decline as gold, the renminbi, and the yen strengthened.

GUY: My inference from Goldman Sachs Exchanges is that this is still a volatility and positioning event until policy confirms it. The yen was estimated to be about twenty-five percent undervalued, and a currency-hedged Japanese investor was said to earn roughly one hundred basis points more on a ten-year JGB than on a ten-year United States Treasury. Those facts can support repatriation, but only if domestic policy makes that relative value credible.

AVA: Now shift to The Indicator from Planet Money. Mike Strain and Claudia Sahm challenged the habit of calling the economy K-shaped or C-shaped. Treasury Secretary Scott Bessent cited a two percent real-wage gain for the bottom quartile and narrowing wage-growth gaps between continuing workers with high-school and college educations. Sahm countered that the stock of wealth remains extremely uneven: the top one percent own about one-third of United States wealth, while the bottom half owns less than three percent.

GUY: The Indicator's sharper distinction was incumbent versus entrant. Unemployment was cited at four point one percent. People who are employed generally like their jobs, wages remain solid, and consumer spending plus artificial-intelligence investment are tailwinds. But labor-force growth has slowed sharply as immigration declines and the population ages. Job creation is weaker, and getting into employment is approaching recessionary conditions even though the aggregate economy is not in recession.

AVA: That is such an important split. The Indicator had Mike Strain describe an orderly cooling from the overheated twenty twenty-two economy, with inflation as his larger concern. Claudia Sahm saw pressure accumulating on consumers, companies, and labor-market entrants, but did not issue a recession warning. So a low unemployment rate is not enough. Hiring rates, labor-force growth, and outcomes for new entrants are the more sensitive indicators.

GUY: The All-In Podcast widened that capacity debate. David Friedberg framed federal arithmetic as roughly seven trillion dollars of annual spending, five trillion of revenue, forty trillion of debt, and a possible two-trillion-dollar interest burden if the long-term rate is five percent. Rahm Emanuel proposed a five-year nominal spending cap alongside revenue reform, criticism of inherited-wealth preferences, and a larger earned-income credit.

AVA: All-In also connected fiscal policy to human capital. Emanuel proposed redirecting some detention spending toward community colleges and skilled trades. He cited a shortage of five hundred thousand to six hundred thousand electricians needed for data centers and one hundred fifty thousand broadband workers. His argument was that immigration, education, energy modernization, and research funding are not separate topics. They form one competitiveness system.

GUY: My inference from The Indicator and All-In is that deficits and artificial-intelligence investment are tolerable when they build productive capacity. They become more inflationary and less credible when labor supply, training, infrastructure, and financing fail to expand with demand. The market may keep celebrating announced capital expenditure, but the bottleneck could migrate into commissioning time, construction wages, and utilization.

AVA: That takes us to technology. On Morgan Stanley's Thoughts on the Market, Andrew Percoco and Tim Hsiao framed robotaxis as a potential one-trillion-dollar market by twenty forty. Their United States forecast rises from one hundred sixteen million autonomous miles in twenty twenty-five to sixteen billion in twenty thirty-two. That sounds huge, yet it would still represent only about zero point five percent of total miles.

GUY: Thoughts on the Market said the profit pool can still be large at about two dollars per mile, but only if utilization is high enough to amortize depreciation and if safety performance reduces insurance, currently estimated near thirty cents per mile. That is the crucial shift: autonomy is becoming an operations business. Another successful demo does not tell you whether the fleet makes money.

AVA: Thoughts on the Market gave four reasons this cycle differs from the autonomy enthusiasm of twenty eighteen and twenty twenty-one: faster improvement in end-to-end artificial intelligence, lower hardware and training costs, better-capitalized operators, and clearer regulation. China provides an operating example with more than five thousand vehicles across over seven thousand five hundred square kilometers in major cities. Some operators average more than twenty orders per vehicle per day.

GUY: And Thoughts on the Market put numbers on the flywheel. Total cost of ownership has fallen roughly thirty to forty percent. Remote assistance has improved from one operator for every twenty to forty vehicles toward fifty to sixty, with one operator per one hundred vehicles viewed as achievable. The vehicle itself was cited around thirty-five thousand to forty-five thousand dollars for a purpose-built Chinese robotaxi, versus about one hundred fifty thousand dollars for a United States vehicle.

AVA: Lower vehicle cost reduces depreciation and therefore lowers the utilization required to break even. But Thoughts on the Market warned that registration, data localization, insurance, and local operating costs can delay the margin curve. Europe, the Middle East, and Southeast Asia together have about four million taxis and ride-hailing vehicles. Twenty-five percent conversion would imply one million Level Four vehicles, with room for regional winners rather than one global monopoly.

GUY: Thoughts on the Market also broadened the value chain. Recurring software subscriptions could reduce automaker cyclicality, while rental-car operators could handle charging, maintenance, reconditioning, and fleet management. But the scorecard stays operational: city rollout, fleet density, paid rides per vehicle, insurance per mile, remote-assistance ratios, vehicle depreciation, and deadhead time. Those measures have to improve before accepting the episode's scenario of margins above thirty percent at scale.

AVA: Now compare that with the a16z Podcast. Joe Schmidt and Andy McCall said enterprise artificial-intelligence companies must choose between two sales systems: lighthouse and land-grab. Lighthouse selling fits a new, regulated, or high-risk category where a credible customer's proof travels to the next buyer. Harvey in legal artificial intelligence and Further AI in insurance were given as examples.

GUY: The a16z Podcast said land-grab selling fits an existing budget and a workflow whose replacement can be justified mathematically. Accounts receivable, customer support, and mid-market operational software can win through higher collections, lower cost, or faster deployment. The labels are not marketing decoration. They determine sales-cycle length, product configuration, customer-acquisition cost, and feedback speed.

AVA: The a16z Podcast used Samsara's electronic-logging-device experience to explain the land-grab flywheel. Regulation created a broad budget, mid-market customers accepted a young vendor, shorter sales cycles accelerated product feedback, and the company later moved up-market and verticalized. Meraki made cloud management tangible with a free access-point trial. The lesson for artificial intelligence is stricter because a proof of concept can easily become an endless science project.

GUY: Exactly. The a16z Podcast said the buyer and vendor should agree in advance on the pilot's duration, success metric, and conversion condition. Artificial intelligence may reopen platform selling because it changes who does the work, not merely the software interface. But the opportunity comes with implementation risk. A prestigious logo that never converts is not proof of a scalable business. Repeatable sales, sound unit economics, and early sales operations matter more.

AVA: My inference from the a16z Podcast is that time-to-value and proof portability are the real go-to-market metrics. If a deployment cannot show a bounded process, a measurable benefit, and a recurring contract, then the vendor may be capitalizing custom services while describing them as software. Better models do not solve a confused sales motion.

GUY: The Vergecast brought this down to the backyard. Jennifer Pattison Tuohy and Brandon Doyle found that robot lawnmowers have improved materially by combining LiDAR, cameras, network RTK, and all-wheel drive. Current products can navigate many ordinary yards with trees, slopes, and narrow passages, and the reviewers estimated that the best products deliver roughly ninety-five percent of the desired result.

AVA: But The Vergecast made the last five percent sound expensive. The remaining problems include obstacle rescues, edge trimming, blade safety, connectivity, and finding a place to dock a large machine. Leading products were discussed around a reasonable sale target of roughly twenty-five hundred dollars, with smaller-yard devices below about eleven hundred to twelve hundred dollars. Network RTK removes a local antenna, but adds subscription and cloud-continuity risk.

GUY: The Vergecast also warned that a new United States FCC restriction on mobile robots could constrain future imports without waivers. So a technically capable device can still fail commercially because of regulation or service continuity. The broader conclusion was that purpose-built robots are useful today, while a reliable general household robot that can manipulate many different objects remains much further away.

AVA: TBPN's official show notes added a partial-source view, and we should underline that these were notes rather than a full transcript. Igor Babuschkin discussed personalized, user-owned models, video games as benchmarks, real-world reinforcement learning, specialized models, GPU demand, and custom inference. Sonya Huang argued that application companies can increasingly own and customize intelligence as post-training tools mature.

GUY: TBPN's official notes also had CoreWeave co-founder Brannin McBee discuss financing, data-center capacity, hardware longevity, and global expansion. Garrett Langley addressed Flock Safety audits and shorter data retention after misuse concerns. Sean Cole described lab-grown neurons doing basic token prediction with nearer-term potential in drug testing. Those are topic-level signals only; the written brief did not infer unprovided guest claims.

AVA: No Priors was also limited to official detailed show notes. Those notes described Chess.com as having ten million daily active users and two hundred fifty million registered members. CEO Erik Allebest discussed the platform's growth since a two-thousand-five domain purchase, the role of General Atlantic and CVC, and uses of artificial intelligence inside the company and product. We should not infer his AGI forecast or roadmap beyond those stated topics.

GUY: My inference from No Priors, TBPN, and the a16z Podcast is that cheaper intelligence can raise the value of distribution, trust, and community. Chess remains culturally valuable after machines became superhuman because the platform owns human identity, competition, and access. Application companies may customize more of the intelligence, but durable value still requires proprietary feedback, governance, and a reason for customers to stay.

AVA: Now geopolitics. On the All-In Podcast, Rahm Emanuel proposed an allied economic bloc linking the United States with Japan, South Korea, Taiwan, Australia, Europe, and willing partners in Latin America and the Middle East. His argument was that China exports domestic overcapacity and uses market dependence as leverage, while political and tariff disputes risk weakening the American coalition.

GUY: All-In had Emanuel call Japan the long pole of Indo-Pacific strategy because it anchors United States force posture, investment, the Quad, and trilateral arrangements with South Korea and the Philippines. On Taiwan, he described three possible coercive paths: economic quarantine, rapid invasion, or seizure of smaller offshore islands followed by a test of whether allies are willing to escalate.

AVA: All-In said Emanuel views pressure on the Philippines and in the South China Sea as the nearer-term deterrence test. His proposed response was more allied military presence and an economic coalition that isolates China rather than the United States. He also endorsed the bipartisan Dignity Act, more capacity for foreign students at American universities, dual-credit high schools, community colleges, and skilled trades.

GUY: The connection to technology in All-In is direct. Data centers, broadband, shipbuilding, defense, and research cannot scale if immigration and training policy leave the labor unavailable. And the connection to Goldman Sachs Exchanges is financial: supporting yen stability helps protect Treasury-market functioning and gives a central ally access to dollar liquidity without forcing disorderly asset sales.

AVA: My inference from All-In and Goldman Sachs Exchanges is that strategic alignment and financial stability reinforce each other only if Japan can normalize monetary policy without detonating global carry trades or its own fiscal position. A stable, stronger Japan adds alliance capacity. A disorderly repricing tightens global financial conditions.

GUY: Let us pull together the cross-currents. First, Goldman Sachs Exchanges and All-In imply a causal chain: weak yen creates imported inflation and political stress in Japan; intervention uses dollar liquidity; that can unwind carry positions and raise Treasury volatility; the pressure then shifts to a Bank of Japan hike and possible domestic repatriation. Intervention is the bridge, not the destination.

AVA: Second, The Indicator, All-In, TBPN, and Thoughts on the Market all point to artificial-intelligence capital expenditure outrunning the operating systems needed to use it. The limiting factor may become skilled labor, fleet operations, maintenance, financing, regulation, or organizational execution before it becomes chip supply. Watch construction wages, interconnection and commissioning time, and actual utilization, not only announced spending.

GUY: Third, Thoughts on the Market and The Vergecast show that physical artificial intelligence is an incremental-margin story, not merely a total-addressable-market story. Technical competence arrives before full economic autonomy. The last five percent of reliability and the cost of human exception handling determine whether the promised software-like margin appears or disappears.

AVA: Fourth, The Indicator, a16z, TBPN, and Thoughts on the Market show why distributional pressure matters. The sources do not prove net job destruction. They do show that workflow replacement, autonomous vehicles, customized models, and even biological computing will distribute benefits and adjustment costs unevenly. Political support will depend on whether productivity reaches wages, prices, and fiscal capacity before labor-market entry becomes materially harder.

GUY: Time for what we are watching. From Goldman Sachs Exchanges, the September twenty twenty-six Bank of Japan meeting is the first hard catalyst. Does the bank deliver the twenty-five-basis-point hike priced at a sixty-five percent probability? Through year-end, test the roughly forty basis points of expected tightening, Japanese asset repatriation, and renewed intervention risk around dollar-yen one hundred sixty and euro-yen one hundred eighty-seven point five zero.

AVA: From Thoughts on the Market, the next robotaxi operating updates should report city count, fleet density, paid rides per vehicle, insurance cost per mile, remote-assistance ratio, and purpose-built vehicle cost. Those figures test the margin case. Capability headlines without improving utilization, safety economics, and free-cash-flow conversion are not enough.

GUY: From the a16z Podcast, watch the next enterprise-artificial-intelligence buying cycle for conversion rates and duration. Do lighthouse deployments produce proof that travels? Do land-grab pilots have fixed success metrics and turn into recurring contracts? A pilot that expands indefinitely is a warning, even if the customer logo looks impressive.

AVA: From TBPN's official notes, watch CoreWeave's next reporting and financing events for contracted capacity, hardware useful-life assumptions, global buildout, and free-cash-flow conversion, not just demand. From The Vergecast, watch implementation of the United States mobile-robot import restriction, especially waivers and whether product launches slow.

GUY: And from The Indicator, into the twenty twenty-six United States midterms, watch labor-force participation, hiring rates for new entrants, real wages for the bottom quartile, and whether the inequality debate moves from alphabet-shaped labels toward broad affordability. The headline unemployment rate can look calm while the entry door narrows.

AVA: The bottom line from today's eight podcasts is simple. Better technology and bigger spending are inputs. Durable returns require operating density, trained people, trustworthy distribution, measurable customer value, credible policy, and financing that survives the buildout.

GUY: That is Morning Signal for Thursday, August thirteenth. Thanks for joining us.

AVA: We will be back with the next verified brief. Until then, watch the constraints, not just the capability.