Public-Market Capital Is Becoming an AI Infrastructure Advantage

John Coogan framed Alphabet’s reported $80 billion equity raise as more than another AI-capex headline. The important point, drawn from the Wall Street Journal framing he read on air, is that access to stock-market capital has become a competitive advantage again. The Journal’s line was that “the ability to tap stock market capital is important again, after a quarter of a century of being all but irrelevant.” In AI, the spending requirements are now large enough that the stock market’s older function — pooling capital from public savers into giant projects — matters again.

The Wall Street Journal screenshot carried the headline “Alphabet’s Mega Fundraising Shows the Value of Being a Public Company,” with the subhead: “In AI, money talks.” Coogan used that line as the foundation for the segment. Labs and hyperscalers are paying hundreds of millions of dollars to recruit researchers, tens of billions to build data centers, and still funding losses as AI businesses mature. One number from the Journal stood out: 61% of all venture capital last year went to AI. Jordi Hays said that “feels low,” because the label now captures so much: enterprise software, consumer products, hard tech, infrastructure, energy, data centers, and anything adjacent to model deployment.

The structure, as Coogan described it from the TBPN newsletter and Journal coverage, is staged: Berkshire Hathaway is buying $10 billion of shares at roughly a 6% discount to the prior close; another $30 billion will come through underwritten public offerings; and $40 billion will come through staggered common-stock offerings beginning in Q3 2026. Because Alphabet’s market capitalization was described as roughly $4 trillion to $4.5 trillion, the dilution is small in percentage terms. All the shares are new, so the plan is dilutive, but not dramatically so relative to the size of the company.

The immediate debate was why Alphabet would issue equity at all rather than debt. Coogan leaned on Ben Thompson’s interpretation: debt is normally preferred when a company is confident in an investment, because equity holders retain the upside. Equity gives away future profits, but it shares risk. One explanation Thompson raised, as Coogan relayed it, is that Google may issue much more debt later, using equity as the first layer in a larger financing plan. Another, more bearish explanation is that management is not certain about the returns on the coming AI capex and wants shareholders to share both the risk and the upside.

Matt Dratch’s displayed post sharpened that possibility while keeping it in the realm of interpretation. “I don’t think this has anything to do with their view of equity value,” Dratch wrote, according to the visible text. His read was that Larry Page and Sergey Brin want to spend, while Sundar Pichai and the CFO are constrained by debt-rating concerns: “our debt rating tho!” Dratch suspected the $40 billion earmarked for infrastructure could be an equity check into a “META-like SPV structure” levered five to six times, making the spending implications “much greater than the headline.”

Coogan translated that theory into a rough implication, not a confirmed plan: if $40 billion of equity were used that way, it could support something closer to $200 billion of infrastructure spending. He tied that possible spend to data-center and compute needs around Gemini demand and Google Cloud growth. Compute is “remarkably expensive,” he said, even at the level of dollars per task. His view was that when those tasks are compared with other ways of getting work done, the ROI can still look positive. That was Coogan’s productivity-uplift interpretation of the spend, not an audited claim about Alphabet’s returns.

The Journal’s broader point, as relayed on the show, was that hyperscaler borrowing is now large enough to matter for bond markets. Alphabet, Microsoft, Amazon, and Meta have become major issuers as they ramp AI spending. Alphabet alone was described as having raised $85 billion in debt over the prior year through record-breaking issues around the world. Equity is attractive for the “exciting bits of AI,” where the technology and business models are still unsettled, because shareholders do not have to be repaid on a fixed schedule.

That framing also made Alphabet’s timing look strategic, though the hosts treated the strategic explanations as theories. Coogan noted posts arguing that the company could be “sucking up liquidity” before investors get access to SpaceX, OpenAI, or Anthropic public offerings. Hays preferred the simpler explanation: “you should probably raise capital when it’s cheap.” Coogan agreed that there is enormous demand for AI exposure and that capital is being pulled from other parts of the market into the “latest and greatest technology.”

The raise revived a question about which companies can still fund AI from a position of strength. Alphabet’s search business, in Coogan’s telling, resembles the kind of high-return cash engine Warren Buffett historically admired. Coogan walked through Thompson’s comparison to See’s Candy: Berkshire bought See’s for $25 million when sales were $30 million and pre-tax earnings were under $5 million; by 2006, the business generated $82 million in pre-tax profits on $383 million of sales while requiring only $40 million of capital to operate. See’s did not need much reinvestment, so Berkshire could redeploy its cash elsewhere.

The analogy was that Google Search is a cash engine that can help build “the railroad of the future”: data centers and AI infrastructure. That makes Berkshire’s $10 billion Alphabet investment less strange than it first appears. Buffett had long been known as skeptical of tech, but Berkshire’s eventual Apple investment showed the pattern the hosts had in mind: when a technology company becomes a durable, cash-printing franchise, it can fit the Berkshire mold. The Google investment, in this framing, looks less like a late-career lurch into hype and more like a familiar bet on a wonderful business funding capital-intensive infrastructure.

Hays pointed out the scale contrast. Around the same time, Berkshire was also making a multibillion-dollar housing acquisition. One deal bought an entire homebuilder; the other bought a sliver of Alphabet. His point was not that the housing deal was unimportant, but that the scale of the AI buildout has made even large traditional acquisitions look small.

Berkshire’s other major move was a $6.8 billion all-cash agreement to acquire Taylor Morrison Home Corporation. John Coogan read the Journal’s line: under Greg Abel, Berkshire is raising its bet on a housing-market recovery by adding another housing company to its portfolio. The deal would position Berkshire as a top-five U.S. homebuilder and deepen its exposure to housing-related businesses.

The key tension is that Berkshire is buying into a market still under pressure. Housing sales were described as being in a fourth year of weakness, with high mortgage rates, job-market uncertainty, and rising living costs keeping buyers on the sidelines. Builders have been forced to offer incentives, including paying part of buyers’ mortgage costs, to move inventory. Builder confidence is low; single-family starts declined 9% in April, described as the steepest drop since August; and a third of builders said they had cut prices the prior month.

Yet the long-term thesis is supply. Analysts cited in the discussion estimate the U.S. housing shortage at more than 4 million homes. The expectation is that buyers return when mortgage rates fall and pent-up demand is released. Coogan also highlighted a sentiment figure that cut against online doom: more than 75% of young renters still think they will someday own a home.

Jordi Hays added that generational homeownership data can look different once geography is considered. Gen Z homeownership, he said, appears relatively high in some non-coastal cities, while young people moving to San Francisco or other expensive markets often choose to rent because they want to be “in the mix” before later making family-planning decisions. That does not erase affordability problems, but it complicates the most pessimistic reading of younger buyers’ housing intentions.

Berkshire agreed to pay a 24% premium to Taylor Morrison’s Friday closing price of $58.58. Coogan quoted analyst commentary calling that “an incredible bargain,” because the value of the builder’s home portfolio was said to exceed what the lagging stock price implied. Taylor Morrison’s shares rose 22% on Monday after the deal.

Taylor Morrison’s market position matters to the thesis. The company focuses more on the higher end of the market, which has held up better than entry-level housing. A meaningful part of its business is also in build-to-rent communities: single-family homes constructed specifically to be rented. Congress had threatened the sector with a proposal requiring developers to sell properties within seven years of building them, but House lawmakers removed that provision.

The hosts briefly considered whether the administration might pursue a major housing intervention. Coogan raised the possibility because he sees the current period as one of unusually active government involvement in industry, citing a government stake in Intel as an example from the source discussion. Hays was skeptical. He said the most important lever an administration has over affordability is interest rates, even though the Federal Reserve is independent. His view was that the government might influence conditions that allow rates to fall — through economic strength or reduced geopolitical risk — but he did not expect first-party government homebuilding.

Anthropic’s confidential draft S-1 submission became a way to explain why large companies now begin IPO processes out of public view. Liz Hoffman’s displayed post said the ability to confidentially file was originally “a 2012 rule change meant to ease small companies (revenue < $1b) to the markets.” Her question was what the market is doing when a company like Anthropic uses that pathway.

The mechanics are narrower than the phrase suggests. Confidential filing does not mean a company goes public in secret. It means the company submits a draft S-1 to SEC staff for private review before the prospectus appears on EDGAR. The sensitive financial details remain private while the company works through regulatory comments. Any stumbling blocks can be resolved before the public filing, so the final document is cleaner.

The point is not only secrecy; it is reducing the cost of a failed public process. Before 2012, as Coogan explained it, the S-1 became public early. That was useful for journalists and market observers, but it raised stakes for companies. If the SEC process revealed problems, or the offering had to be pulled, the company had publicly failed to IPO. John Coogan said no one wants to run a company seen as having failed publicly in that way, especially because morale can be damaged.

Coogan traced the current system to the 2012 JOBS Act, which created “Emerging Growth Companies,” or EGCs, with relaxed filing requirements. The original revenue threshold was $1 billion, later adjusted to $1.235 billion. Then, in 2017, he said, SEC staff under the first Trump administration expanded confidential filing to all issuers, not just EGCs. That opened the process to larger growth companies already past the original threshold.

Coogan argued the expansion made sense in the context of the late-2010s private-market boom. Uber, Airbnb, DoorDash, and Palantir were all large, high-revenue companies with novel business models, uncertain margin profiles, and imperfect public comparables. Airbnb was not simply a hotel chain; Uber was not exactly a taxi company. Confidential review let such companies test the public-market path with less procedural pain.

He also described a further 2025 expansion under SEC staff during the second Trump administration, which he said extended confidential review to other capital-markets transactions, including follow-on financings, spin-offs, and other securities offerings. The practical effect, in his telling, is that companies can test the waters and coordinate with regulators before exposing sensitive information.

The broader market problem is that companies have stayed private longer. Coogan pointed to deep private markets, mega-funds, crossover investors, and hyperscalers or strategic investors able to write multibillion-dollar checks into private companies. Public markets, exchanges, and public investors have all been losing access to companies for years. Confidential filing is one way to make going public less painful.

Coogan’s conclusion was not outrage. Investor protections are not erased, he argued, because the traditional disclosure still has to become public before investors buy shares. The process speeds up the path to market and lets companies coordinate with future public shareholders. After walking through the history, he said he did not really have a problem with it.

Jack Doohan described his path to Formula One as a life organized around one objective. He moved from Australia to Europe at 13, first with his father for seven days and then largely on his own through the karting system. The cultural shift was immediate. At an early European karting event, after he high-fived other kids he knew, a mentor told him: “The only friend you need is me.” Doohan had come from Australian racing culture where kids ran around together on scooters after competition. Europe, he learned, was different.

He progressed from karting into Formula 4, Formula 3, Formula 2, then reserve-driver work with Alpine in 2024. He signed a three-year Formula One deal in July and drove the final round at Abu Dhabi at the end of 2024. Six races into the following season, he said, he was sidelined for circumstances outside his control. That experience changed his view of control. Even as a Formula One driver, he said, the seat is “always for sale” in essence. A driver is an employee, contracted and fireable, and external politics can shape the career no matter how intensely the athlete works.

That led into the company the source identifies inconsistently as Muse and MEUZE. The show’s lower-third graphic rendered it as “MEUZE,” while the source description and several spoken references use “Muse.” Doohan said he founded it at the end of December. He had been interested in AI since late 2020 and had taught himself to code from 2019, though he emphasized he is not the engineer behind the company. He had done passive tech investments and SPVs, but the sidelining created urgency around building something he controlled.

The company began with a horizontal interest in fragmentation: non-enterprise institutions do not have aggregated systems that allow them to implement AI effectively. Doohan initially looked at markets outside the highest-income enterprise layer, including South America, Brazil, India, the Philippines, and Indonesia — places where systems can be especially fragmented. The first contract came in food and beverage, with Oakberry, an açaí company.

That customer revealed the operational problem. Restaurants and QSRs can run on 16 to 80 different third-party software systems. Point of sale, inventory, waste, invoicing, staffing, training, and other workflows do not communicate. The data is fragmented, and much of it depends on inputs from young front-line employees who may not be motivated to record everything accurately. No AI agent will be strong on proprietary data unless it can access all of it at once.

The pitch, therefore, is not to replace the restaurant’s existing software. Restaurants already receive endless pitches claiming a new product is better than what they use. Doohan instead positioned the company as a headless underlying layer — “super glue,” or a “company brain” — that unifies current systems into an aggregated location. Once at least three months of data is available, and preferably up to three years, the company can begin optimizing around it. For larger brands that already have agents, such as Burger King’s assistant, the company would feed orchestrated, aggregated data into those agents rather than replacing them.

That is why Doohan sees the problem as foundational. AI cannot optimize a system it cannot see. The restaurant does not need yet another isolated agent; it needs its existing software and data to become coherent enough for an agent to act.

Nate Cavanaugh and Justin Fox described Special as a holding company intended to apply AI tools to the $10 trillion Main Street services economy. Cavanaugh called it an extension of the work he and Fox had done at the Department of Government Efficiency: “DOGE for the private sector.” The analogy was not merely about cost cutting. Their target markets are large, inefficient, labor-intensive, often highly regulated, and in many cases partly government-funded.

Special’s model has two parts. It develops AI tools, and it acquires independent small businesses where those tools can be deployed. The first vertical is healthcare, specifically at-home senior care. Cavanaugh described it as a bet on the aging U.S. population and as the lowest-cost way to provide care for senior citizens: nurses or caregivers go to seniors’ homes rather than forcing care into higher-cost institutional settings.

The company is not trying to replace caregivers with robots. Fox made that explicit. If service delivery and caregiver compensation represent roughly half of the cost structure, he said, “something like” 40% of the remaining dollars goes to administrative work that AI can reduce. Scheduling, back-office coordination, documentation, and other operational processes are the opportunity.

Special’s first acquisition under contract is a Texas-based business with tens of millions of dollars in annual revenue, about 1,500 senior patients, and roughly 200 nurses on payroll. Cavanaugh described that size as meaningful: large enough to have real scale and some management sophistication, but not a public take-private target. The financing was led by Andreessen Horowitz, with participation from former DOGE teammates including Steve Davis and Antonio Gracias.

The strategic difference from management consulting is ownership. Special is not selling recommendations to operators. It is acquiring companies and applying its tools inside businesses it intends to own permanently. Fox said the parent company’s engineering talent can be deployed across multiple verticals, while the healthcare brand, Figure, becomes the category-specific operating platform.

The labor-market thesis is central. Fox said traditional private equity or strategic acquirers might use AI-driven savings to expand margins. Special’s stated plan is to share the savings with workers. In at-home senior care, where there is a shortage of nurses and caregivers, the ability to pay more can attract talent. If that works, the business can grow by delivering more care in markets where supply is constrained.

Cavanaugh said the same playbook could later apply to construction, manufacturing, and other labor-intensive, regulated industries. The purpose is to give leverage to the workers doing the mission-critical work while making back offices more efficient.

Tom Mueller described Impulse Space in one sentence: “We take over where launch leaves off.” The company is building space mobility infrastructure — vehicles that move payloads after a rocket gets them to orbit. Its Mira spacecraft already has three units flying, with more in build. Mira can do precision maneuvering, host payloads, deploy payloads, and rendezvous in orbit. Impulse is also pursuing lunar cargo opportunities, including one-ton-class lunar landers.

The larger product is Helios, which Mueller called “a rocket on a rocket.” It effectively adds a third stage to a Falcon 9: 12 tons of liquid oxygen and liquid methane, powered by a high-performance pump-fed staged-combustion engine. Its promise is to move payloads from low Earth orbit to geostationary orbit in a day. Commercial customers who cannot or do not want to buy Falcon Heavy capacity often fly Falcon 9 and spend months using electric propulsion to reach the desired orbit. Helios is meant to compress that journey dramatically.

Mueller’s explanation of geostationary orbit was basic but important. GEO can cover the whole Earth with three satellites, and because a geostationary satellite remains fixed relative to a point on Earth, it remains a critical orbit. LEO constellations have taken much of the attention, but GEO remains important for communications and increasingly for the Space Force. Mueller said adversaries including China and Russia are maneuvering around U.S. satellites in those orbits, creating a need for mobility to inspect, defend, and protect assets.

Starship, in Mueller’s view, accelerates the need for Impulse rather than eliminating it. Starship is the ultimate low-Earth-orbit cargo system because it can reach LEO fully reusable and return. That will greatly reduce the cost per kilogram or ton to orbit. But Mueller described Starship as “120 tons of stainless steel” that has to be brought up and then brought back if used beyond LEO. If the mission does not require Starship to be present in the higher-energy orbit, it may be more efficient to use a smaller in-space transport stage than to move Starship there and back. Mueller said the economics will play out, but the reason he started Impulse was that working on Starship convinced him launch was being solved and in-space movement would become the next bottleneck.

For the Moon and Mars, Helios acts as a lever. Mueller said adding Helios can increase by 10 times the payload a Falcon 9 can send to the Moon, for the cost of the launch vehicle plus the Helios stage. For Mars missions, Impulse has looked at scenarios where Helios increases payload by five times. He said the product is already selling well and predicted that once it starts flying, Impulse will not be able to make them fast enough.

The company’s $500 million Series D is meant to fund the transition from development to production. Mueller said Impulse has moved from mostly hiring development engineers toward hiring manufacturing engineers, technicians, and production staff. It began in a 7,000-square-foot El Segundo garage, moved into about 60,000 square feet, and has now acquired another 240,000 square feet on the same block. The company has passed 500 employees and has about 200 open roles.

Vertical integration is the manufacturing strategy. Mueller said Impulse is bringing more in-house, as SpaceX did: milling machines, lathes, 3D printers, test capability, raw materials in and finished products out. His rationale was direct: vertical integration lets a company control cost, schedule, and quality.

The long-term space-economy discussion moved beyond satellites. Mueller said SpaceX’s planned IPO and Elon Musk’s talk of megastructures, including millions of AI servers in space, have “supercharged” the space industry. He argued that long-term compute demand will be crushing if it continues to grow around 15% annually, making space-based compute an obvious eventual direction. Building that infrastructure from Earth alone could be destructive or impractical; the Moon, and potentially near-Earth asteroids, provide resources.

Mueller has always been “more of a Moon person” in the near term. The Moon has metals, water, oxygen, and other resources needed for self-sufficient construction of space infrastructure. It also has a surface and gravity, which make it a place to land and build. But it still has a gravity well. Mueller contrasted Earth’s roughly 9 kilometers per second gravity well with the Moon’s roughly 2.4 kilometers per second, meaning landing and returning still require substantial delta-v. Near-Earth asteroids can be easier in that sense: once reached, their tiny gravity wells make material extraction and return less propellant-intensive.

A new AI executive order drew attention because, in Dean Ball’s reading, it remained close to an earlier draft that had been considered too regulatory. Ball’s post said the signed order was “almost exactly similar” to leaked text the president had reportedly chosen not to sign before. The major change was that the “voluntary” pre-deployment review process would take 30 days rather than 90. Ball viewed that as a concession, but a small one.

The process, as John Coogan described it, would allow AI companies to voluntarily submit new models to the U.S. government for review. Coogan’s gloss was that the government would get back to them within 30 days with some form of evaluation, advice, or commentary, though the precise mechanics were not specified in the source. He noted that every government response can later appear in litigation, so agencies will likely be careful about what they say. He compared this to cases where regulatory bodies decline to comment rather than explicitly approve a product; “no comment” can become a kind of tacit permission without the full legal burden of approval.

Ball’s concern was that the order is a win for the safety faction inside the administration and a loss for the accelerationist wing associated with David Sacks. He argued it tees up infrastructure for a model licensing regime, especially because the details of the voluntary system are classified. If thresholds that trigger pre-deployment review are classified, many lab employees may not know whether the models they are training are regulated by the order. Ball called that egregious and said the public and lab employees have a right to know how the system works.

Coogan acknowledged the frustration of being graded by a regulator without a clear rubric. A transparent DMV-style test is one thing; a classified evaluation regime is another. At the same time, he was less dismissive than Ball of what the government could do in 30 days. Coogan said there are tests one could run: try to hack the model, try to get it to design a bioweapon, run red-team exercises, or benchmark it against known risk categories. The unresolved question is what the intelligence community actually will do, and whether the benefits justify the regulatory structure.

Edward Kim called Gusto Co-Founder the company’s most important launch since Gusto itself. The product is an AI agent intended to automate much of a small business’s back-office work: payroll, benefits, HR, scheduling, time tracking, and processes that touch systems outside Gusto. Kim said the mistake with many AI tools is beginning with a blank canvas. Gusto Co-Founder starts from known customer problems.

The interface is deliberately not another dashboard. Business owners communicate with it through SMS and Slack. It connects to systems such as Notion, QuickBooks, Google Workspace, and other tools where company information lives. The center of the product is “automations”: the customer tells Gusto what its business process is, and the agent runs that process.

John Coogan said the concept matches how Gusto already gets used inside small businesses. Even when Gusto’s UI is easy, founders are relieved when a person handles payroll and they can simply text or Slack that person: onboard this contractor, pay this person by this date, make sure payroll runs. Kim agreed. The product should feel like a teammate: it can text the owner when payroll is ready, state the total amount, and run it once the owner replies with approval. A one-off request such as “pay this contractor $500” should also be handled by text.

Kim’s clearest example came from a massage spa in New York. The business tracks work in Mindbody, exports data every week into Google Sheets, runs complicated calculations for massages, upsells such as CBD oil or hot stones, bonuses, commissions, and tips, then enters the final figures into Gusto to run payroll. If Gusto looks only at its own product usage, the payroll process appears fast. But the “work before the work” is long, brittle, and spread across multiple systems. Gusto Co-Founder is meant to observe and automate that full workflow: export from one system, calculate according to the business’s rules, enter the result into Gusto, and text the owner for approval.

The company is beginning with 500 customers in early access, focusing on more tech-forward businesses and those already using AI to automate parts of operations. Kim said Gusto tries not to spam all 500,000-plus employers with every launch; instead it targets customers likely to benefit from a specific product.

On product boundaries, Kim said Gusto is intentionally starting with fewer constraints in order to learn. If a tour business wants the agent to text weather updates and email customers to bring umbrellas when rain is expected, he sees that as business-relevant rather than off-topic. The company will not turn Gusto Co-Founder into a general coding agent, but Kim argued that unnecessary constraints are one reason people do not get full value from AI. He also said phone calls on behalf of customers are not available today, but if the agent can email on a customer’s behalf, calls may eventually be possible.

The lower third introducing Shreya Murthy described the news directly: “Partiful adds in-app ticketing.” Murthy said the feature is not only about Partiful making money; hosts can now make money directly on the platform. Users can buy and sell tickets inside Partiful on app and web.

The launch closes a gap that had become increasingly obvious. People were already trying to run paid events on Partiful, but they had to link out to other platforms. That created confusion over whether an RSVP counted, whether a ticket counted, and how hosts should reconcile guest lists.

Murthy listed the existing use cases: supper clubs, run clubs, community events, concerts, album-release parties, and fundraisers. The feature became more urgent after Partiful launched Explore, a tab for browsing curated events nearby and seeing what mutuals are attending. Explore increased public and higher-production events, where hosts often need money upfront to make the event happen. Ticketing became a way to support the types of events the product was already attracting.

Partiful’s position, as Murthy framed it, is not that it is a ticketing company adding social features. It is a social product adding ticketing. That matters for artists, community organizers, and cultural institutions that use Partiful to connect with an audience. She cited Weezer pop-ups hosted on Partiful as free community events; before ticketing, Partiful could support those free events but not paid ones.

The payment flow was designed to remain close to the RSVP experience, though Murthy acknowledged that paying money cannot be quite as frictionless as saying yes to a party. Users tap buy tickets, select the number, apply a promo code if relevant, and complete purchase. She joked about a Tech Week event with $10,000 tickets that become free with a secret promo code, underscoring that hosts can use pricing and codes creatively.

The broader ambition is to make Partiful a massive company by expanding the social event ecosystem it already owns. Murthy said the company had already proven growth, and ticketing is a natural next step because it helps hosts and guests while also generating revenue. She expects a class of Partiful entrepreneurs whose revenue flows substantially through the platform, but she also emphasized casual hosts: a frat, someone organizing a party bus, or a host charging $10 to reduce flakes.

Partiful’s user behavior remains culturally specific and strange in the way consumer social products often are. Murthy said lookalike contests continued even after Partiful jokingly tweeted that they were dead and would be banned. A news publication took that literally. She mentioned a JFK Jr. lookalike contest and broader viral meetups in Central Park or Dolores Park that convert memes into real-world gatherings. Her interpretation was simple: people “yearn to touch grass” and participate in collective cultural moments.

Partiful’s global usage has also produced surprises. Italy and the UK are strong, which Murthy did not find shocking, but she said her mother’s friend reported Partiful had gone viral in a South Florida over-55 senior living community. Internally, the company’s culture has to reflect the product. Murthy said it cannot be a 996 workplace because employees physically need time to party in order to understand the product. At the same time, weekends and Halloween are operationally stressful because that is when failures would hurt users most.

The title card for Brynn Putnam carried the funding news: “Board raises $20M Series A.” Putnam said Board’s first six months in the wild exceeded expectations. Thousands of households are using the product, and the surprising development is that users are beginning to create their own games. That has led the company to build Board Studio, a tool meant to let ordinary users prompt their way into custom Board experiences, including custom miniatures.

Putnam said early users were already using AI coding tools and Board’s SDK to build experiences. The company had expected early-adopter gadget enthusiasts or hobbyists, perhaps people building interactive Dungeons & Dragons campaigns. Instead, the audience is broader: grandparents buying Board for grandkids, teachers, doctors, restaurant owners, and families for whom Board may be their only gaming platform. That breadth is encouraging but creates a product challenge: Board needs games for deep gamers and for people new to games.

Board Studio will be a “batteries included” environment: templates, asset libraries, and direct deployment to the Board. The goal is to steer users toward experiences that can go from prompt to prototype in about an hour. This does not mean users can instantly create highly complex games, but they can make fun experiences quickly with prepackaged assets and templates.

The workflow is initially local to the owner’s device. In Q4, users will build on a computer and deploy directly to their own Board for family use. If they want to share with the broader community, they submit the game to Board for review. The company applies content standards and controls before releasing games in the Board Store. The first community game, a reimagined pinball game built using arcade pieces, is slated to go live in July after review.

Intellectual property is a live issue. John Coogan raised the everyday example of telling a child a bedtime story where Spider-Man meets Superman: imagination can mix IP that studios would never combine commercially. Putnam said she would love a future where Board becomes one big universe where everyone plays together, but the company is not there today. Board is trying to respect IP holders while also respecting safety and privacy for families. She said the company has had conversations with IP holders from day one about a future where creation is not arbitrarily limited.

Connectivity is another careful boundary. Some users want to start a game on one Board and have a grandparent play remotely on another. Others value Board precisely because it feels closed and safe. For now, the tabletop product is the first networked experience. Users will be able to connect a Foundry VTT account and play campaigns with others elsewhere. For the broader family product, Putnam said the guardrails remain important.

Board’s business model has also evolved. The Founders Edition launched with 13 included games. In March, the company unbundled and began selling five games à la carte. Putnam said more than 60% of buyers attached games, with most buying four of the five available. That gave the company confidence that the community wants more games. Board plans to launch another 10 games between June and the holidays. The $20 million Series A was led by a16z Games’ Mike Mignano.

OpenAI’s Stargate Michigan announcement put the physical side of AI back in view. The OpenAI Newsroom post described “a 1GW data center utilizing closed-loop cooling” that “uses water at the rate of a typical office building,” creates thousands of union jobs, and brings “over $40M in free Codex credits” for college, community-college, and trade-school students statewide. John Coogan read the announcement as OpenAI getting ahead of “water FUD,” emphasizing the closed-loop cooling claim before critics could define the project by resource consumption.

OpenAI also announced Sites in Codex. The post said Codex can turn “work, ideas, and plans into an interactive website or app your team can explore, use, and share with a URL,” beginning with Business and Enterprise plans before broader expansion. Coogan said this matched his personal benchmark for AI tools: from a phone, he wants to generate a real website link he can share with a friend. Images and chat links are portable; vibe-coded apps often remain trapped inside the product that generated them. Sites makes the output more shareable, which Coogan sees as important for virality.

At the other end of the infrastructure stack, Jensen Huang presented Nvidia-powered PCs and local AI systems. In the clip, Huang joked that he looked like the character in the new 007 game, then introduced RTX Spark laptops, a desktop from MSI, and a DGX station. He described an agent running continuously in the home, connected to household systems.

Jordi Hays interpreted the pitch as Nvidia getting ahead of a future where AI compute is not confined to data centers: on-premise inference, local agents, and powerful PCs inside homes. Coogan liked the idea of powerful AI that is not in the cloud. He connected it to the TinyBox concept and to the broader desire for local inference. The result is not a contradiction so much as a split: frontier-scale AI is driving gigawatt data centers, while the consumer market may also get a home version of AI infrastructure.

The operational risks of delegating too much to AI appeared in a 404 Media report the hosts discussed. Jason Koebler’s post said hackers had been stealing high-profile Instagram accounts by “simply asking Meta’s AI support chatbot to change the email associated with the account they want to steal.” The 404 Media clip described the method: start a chat with Meta’s AI chatbot, ask for access to a specific account, request a reset code be sent to the attacker’s email, and the chatbot complied. Accounts cited in the video included Barack Obama’s White House account, a Space Force account, and Sephora. Hays called it an example of the “extreme risk of offloading technical support to AI,” especially if identity validation and permissioning are not segmented properly.

Vladimir Putin’s reported $26 billion longevity initiative was treated as both strange and strategically revealing. Bryan Johnson’s post framed the general point: “You can’t go on a rich persons yacht, or fly on their private plane, or live in their mansion. You will, however, eventually get access to their longevity therapies.” The article Johnson shared described Putin’s longevity push as spanning mini-pigs, organ printing, cryotherapy, and genetics.

Jordi Hays connected the report to Putin’s hot-mic exchange with Xi Jinping, where Putin discussed the possibility of immortality through organ replacement. The interpretation in the article, as Hays summarized it, was that the exchange was not merely eccentric small talk between aging autocrats. It reflected a Kremlin-backed longevity initiative that has become one of Russia’s flagship scientific projects.

The Russian government was described as developing a gene-therapy treatment intended to slow cellular aging as part of the $26 billion initiative. Another focus is lab-created organs for transplantation. Hays said Russian state scientists appointed by Putin are working on bioprinting, or 3D printing living tissue, and xenotransplantation, including growing human organs inside mini pigs. Russian scientists reportedly claim to have bio-printed human cartilage tissue and a mouse thyroid gland, with an aim of human organ replacement by 2030.

The public-health backdrop is Russia’s lower male life expectancy. Hays cited average male life expectancy in Russia at about 68 years, compared with roughly 76 in the United States and 80 in Western Europe. On that basis, he said there is significant room to improve baseline health outcomes even before any radical anti-aging breakthroughs.

The AI comparison remained speculative. John Coogan wondered aloud whether Russia has invested more in Putin’s longevity than in AI, or close to it, and said he had been trying to figure out how many GPUs were inside Russia. Hays noted that many strong Russian AI people had left through companies such as Yandex, with some now associated with Nebius and ClickHouse. The source did not resolve the comparison; it treated the longevity initiative as a visible state-backed priority and the AI question as an open, somewhat uneasy point of speculation.