Jeremy Allaire

Roblox’s AI thesis extends a 20-year plan to build a photorealistic, physics-backed “Holodeck” where 10,000 people can modify a shared world in real time. Baszucki sees a continuum from human “coexperience” to solitary “real-time dreaming,” with NPCs at the latter end. If 4D simulation succeeds, “video is the downsampling” and Zoom becomes a “legacy analog mode.”

Baszucki centers the hardest technical problem on synchronizing the state and history of 10,000 participants. Roblox is testing whether that state belongs in video latents, native 3D or a new hybrid format. Its potential advantage is 13 billion hours of monthly interaction data that can be reconstructed from any camera angle rather than stored as fixed raster video.

Roblox’s NPC roadmap progresses from general game-playing agents to opt-in virtual doppelgangers and then agentic substitutes. Baszucki imagines an NPC learning a user’s gestures, the way they look at things and their speech, then briefly playing with that user’s child while the parent works: “Could my virtual doppelganger fit in for 15 minutes?” The discussion also imagines creators prompting characters such as “a happy Ben Franklin.”

Baszucki envisions a hybrid AI architecture, not one model running the entire world. A high-efficiency multiplayer engine could synchronize state while dedicated systems handle NPCs, 3D upsampling and local 2D photorealism. Roblox is “doubling down on a hybrid, multi-AI tech stack that supports multiplayer natively,” even as direct video-generation models offer a possible parallel path toward personalized dream worlds.

Cheaper generation will raise output quality without necessarily making game development less competitive. Baszucki’s counter to the asset-cost thesis is that “the expectation of quality from consumers goes up at exactly the same velocity.” Five-person teams may produce “astounding” work, but users will normalize that standard; agents could meanwhile spend 24 hours testing builds with 20 simulated players across phones and other clients.

On AI relationships, Baszucki answers “zero right now.” The transcript inconsistently attributes the accompanying “I don’t think we’re close to crossing the human-AI barrier” line, but Baszucki is more receptive to an always-present coach or therapist and says mental-health advice and coaching are already major use cases. Sarah calls that arguably valuable.

Management’s intermediate target is 10% of global gaming content, framed as roughly 300 million DAUs, about 20 billion in an unspecified metric and about 3x Roblox’s current position. Baszucki pairs that measurable waypoint with the much larger Holodeck vision: “Knowing both 3x and 10x is very helpful.” The conversation uses Microsoft Excel—still recognizable after 40 years—as a frame for the possibility that a correctly built Roblox could likewise be “kicking around in 40 years.”

Circle’s 2026 thesis is that regulated stablecoins are becoming an internet-scale financial network, not merely a crypto product. Jeremy Allaire describes USDC as an “HTTP for dollars”: programmable money that moves peer-to-peer while retaining reserves, audits and regulatory trust. The opportunity he cites is roughly $120 trillion of legal electronic money, including about $60 trillion in physical cash and non-interest-bearing demand deposits.

Lower interest rates may reduce Circle’s yield per dollar yet accelerate adoption enough to enlarge the business. Allaire says USDC posted “thousand-percent year-over-year growth” for two straight years when rates were very low, then circulation declined as rates rose; since rates fell from roughly 5.25%-5.5% to about 3.5%, circulation has increased by multiple hundreds of percent. His blunt conclusion: “We really need interest rates to come down” because cheaper money increases investment, velocity and stablecoin usage.

USDC’s moat is the combined network of regulation, liquidity and integrations, which makes a net-new dollar stablecoin worth “essentially zero.” Cash App, Coinbase, Revolut, Visa, Stripe, Shopify and global banks add interoperability whenever they adopt USDC; Circle also operates across the EU, Singapore and the UAE. Allaire expects more competition under the GENIUS Act, including potentially from Tether, but rejects the idea that Amazon, every bank or every platform needs its own stablecoin.

AI is minting more capable attackers and compressing attack timelines, forcing cybersecurity itself to become autonomous. CrowdStrike’s George Kurtz says prompt-only malware can adapt uniquely to each victim, work through an LLM and never phone home, eliminating a traditional detection signal. CrowdStrike is extending endpoint detection and response into “AIDR”—AI detection and response—because Kurtz expects each employee eventually to control roughly 90 agents.

Archer expects visible US eVTOL operations in summer 2026, but public demonstrations precede full certification and scaled passenger service. The DOT is expected to select five trial cities in the first quarter; Archer is pushing Los Angeles, where it holds the LA28 Olympics air-taxi rights and paid roughly $170 million for control of Hawthorne Airport and surrounding assets. Adam Goldstein expects deployment to resemble Waymo’s: tens of aircraft first, potentially hundreds over five to 10 years, with trust earned gradually.

Archer’s investment case rests on redundancy, public-market financing and a defense derivative—not immediate autonomous mass transit. The aircraft has 12 rotors, 24 redundant motors, a 50-foot wing and as much as 10 miles of glide; Goldstein says the target is “an order of magnitude safer” than helicopters, while admitting cascading failures remain the scenario engineers must test. Archer has raised about $4 billion, and its Anduril partnership targets an autonomous “attritable” attack aircraft with potentially 90% lower cost than a $50 million-$70 million Apache-class asset.

AI infrastructure demand is accelerating, while power, construction and capital formation constrain supply. Crusoe’s Abilene site combines a 1.2-gigawatt substation, a 350-megawatt gas plant and 8,000 daily workers; its wider pipeline exceeds 45 gigawatts, while rack density is projected to rise from Blackwell’s 130 kW to Vera Rubin’s 250 kW, Vera Rubin Ultra’s 600 kW and ultimately one megawatt. Chase Lochmiller argues even a failed model provider would leave valuable capacity for its winner: “Nobody has enough compute.”