Ideas
Bearish AR, more bullish VR.
John Coogan is bearish on augmented reality because AR glasses only add light and cannot produce dark pixels against bright backgrounds, limiting hologram visibility, and the category lacks a compelling content ecosystem beyond Snapchat lenses. He is more bullish on virtual reality, where full immersion, passthrough cameras, and large virtual screens work better for movies, games, and wide field-of-view experiences, though adoption may remain slow until devices become lighter and cheaper.
Bearish AR, more bullish VR.
John Coogan is bearish on augmented reality because AR glasses only add light and cannot produce dark pixels against bright backgrounds, limiting hologram visibility, and the category lacks a compelling content ecosystem beyond Snapchat lenses. He is more bullish on virtual reality, where full immersion, passthrough cameras, and large virtual screens work better for movies, games, and wide field-of-view experiences, though adoption may remain slow until devices become lighter and cheaper.
Post-AGI unlocks robotics, drugs, simulation.
Aaref Hilaly argues that once intelligence becomes abundant post-AGI, new capabilities will emerge in areas like robotics, drug discovery, simulation platforms, and AI applications. These are not just incremental improvements; they can unlock tasks humans cannot do today, such as robots doing household chores, AI discovering new drugs, and simulation tools for better decision-making. He cites private companies like Sunday Robotics, Chai Discovery, and Simile as early examples.
AI infrastructure needs power, silicon, GPUs.
Hilaly says the clearest post-AGI investment area below the frontier labs is AI infrastructure: power, silicon, and GPUs are necessary for the buildout and demand is obvious. While some investors only want land, copper, power, data centers, and chip fabs, he is comfortable investing in the infrastructure layer because compute and energy needs are real and growing.
Japan PE opportunity is reaccelerating.
Gross sees a renewed investment opportunity in Japan as the country faces competition from China, Korea, and the U.S. and needs to reinvigorate its economy. Bain initially focused on domestic, insulated, inefficient Japanese businesses where it could control macro variables and improve operations, but as Western private equity became more accepted, it moved into larger technology crown jewels like semiconductor and high-tech businesses. He notes talent attitudes have shifted, with Japanese professionals now viewing Western PE as a way to help revive Japan.
Legal AI has effectively infinite demand.
Junestrand argues legal AI is an existential opportunity and threat to the current operating model. Legal work demand is effectively infinite: if AI makes review cheap, companies can review every contract, handle every claim, fight every litigation, and file every patent. He says some work that took 130 units of time can now be done in 1, and that in-house legal teams and law firms are adopting platforms like Legora to move beyond shallow chatbot use.
AI data center demand outstrips power supply.
Lochmiller says demand for AI infrastructure and power has grown so much that each incremental gigawatt is increasingly challenging, making real-world power production and data center facilities key bottlenecks. Crusoe's energy-first strategy and Abilene campus were designed as a gigawatt-scale coherent cluster to train next-generation frontier models; buildings 3-8 will quadruple compute capacity, while the Spark product serves smaller distributed inference clusters to reduce time-to-token.
Solar and batteries power AI factories.
Lochmiller highlights solar plus battery storage as a compelling way to power AI data centers. Crusoe, with Redwood Materials, built the largest microgrid in the United States, using mostly second-life EV lithium-ion batteries, and those AI factories are powered 99%+ by the sun. This supports the idea that solar and battery storage can unlock AI compute in locations with abundant renewable energy.
SMR nuclear can power AI factories.
Crusoe announced a partnership to bring the first SMR-powered AI factory online by mid-2027. Lochmiller frames small modular nuclear as part of a broader set of new energy technologies that can provide modular, scalable power for AI data centers, especially where gigawatt-scale grid power is hard to secure.
AI customer service agents scaling globally.
Zhang argues that AI customer-service agents are ready to replace frustrating phone trees and legacy support systems, but deployment requires mapping complex enterprise workflows, similar to drawing the map for a self-driving car. Decagon is expanding internationally to Brazil after seeing strong demand from large enterprises and is optimizing model routing, smaller models, and latency to make voice agents cost-effective and snappy.
AI safety/trust remains an ongoing need.
Leonard argues AI safety is not a solved problem; as capabilities advance, the operational definition of safe behavior changes, so guardrails, red teaming, observability, and evals remain ongoing needs. He believes AI should be aligned to the values of real-economy businesses and their customers, not just the frontier labs, and that Hays Labs can provide AI that enterprises can depend on as they adopt it.
SMB AI adoption gap is the opportunity.
Pujji argues the real economy is far behind on AI adoption: SMBs are only about 20% adopted versus 85% of the Fortune 500, and the main barrier is trust, not access. Beacon acquires decades-old software companies serving real-economy businesses like campgrounds, construction, and labor unions, then makes them AI-native. By combining Beacon's trusted customer relationships with Hays Labs' safety and guardrails, it aims to close the adoption gap and deliver AI abundance to the real economy.
This TBPN video, published September 17, 2026,
features John Coogan, Aaref Hilaly, David Gross, Max Junestrand, Chase Lochmiller, Jesse Zhang, Leonard, Jesse Pujji
discussing Augmented reality, Virtual reality, ROBO, Simulation platforms, AI-SECTOR, AI infrastructure supply chain, EWJ, Legal AI, AI Data Center Infrastructure, SOLAR, Battery storage, URA, AI safety/trust infrastructure.
12 trade ideas extracted by AI with direction and confidence scoring.
Speakers:
John Coogan,
Aaref Hilaly,
David Gross,
Max Junestrand,
Chase Lochmiller,
Jesse Zhang,
Leonard,
Jesse Pujji
· Tickers:
Augmented reality,
Virtual reality,
ROBO,
Simulation platforms,
AI-SECTOR,
AI infrastructure supply chain,
EWJ,
Legal AI,
AI Data Center Infrastructure,
SOLAR,
Battery storage,
URA,
AI safety/trust infrastructure
