May 2026
Vincent Koc of Comet ML argues that AI evaluation is being outpaced by the systems it is meant to measure. In a talk on adaptive evaluation for agents, Koc says static benchmarks and handcrafted test sets are poorly suited to applications that change with prompts, tools, production traces, user behavior and even their own harnesses. His proposed direction is to define the intended end state, use traces and telemetry to surface drift and edge cases, and treat evals as a continuously revised system rather than a one-time benchmark.
Public arguments about birth rates quickly become fights over feminism, coercion and political identity, Chris Williamson, Lyman Stone, Stephen J. Shaw and Simone Collins argue in a Modern Wisdom discussion. Stone says the backlash is rooted in a real tension between gender egalitarianism and the cultural patterns associated with higher fertility; Shaw wants the issue framed around helping people have the children they already want, without pressure. Collins is willing to let hostile subcultures opt out of the future, while Williamson argues that abandoning the left makes any effort to raise birth rates too narrow.
In a TED talk, creator Maya Higa argues that conservation education can be built for the internet rather than adapted to it. She presents Alveus Sanctuary, her Austin-based rescue facility with 36 livestreaming cameras and no public visitors, as a model for reaching young audiences where they already spend time while reducing stress on animals. Higa’s case is that sustained online attachment to individual rescued animals can translate into donations, awareness of wildlife pressures and a broader conservation constituency.
Lloyd Blankfein, the former Goldman Sachs chief executive, argues in a conversation with a16z’s David Haber that resilient institutions are built less on prediction than on disciplined contingency planning. Drawing on Goldman’s partnership culture, its financial-crisis risk controls and his view of AI, Blankfein says leaders must take risk while preserving the systems, information flow and judgment needed to survive being wrong.
Anti-Muslim politics in the US and UK works by recasting Islam not as a religion but as an ideology, racial threat or civilizational enemy, according to legal scholar Asma Uddin, journalist Hannah Allam and British commentator Fraser Nelson. Uddin argues that this move can push Muslims outside religious-liberty protections; Nelson sees it as a revival of sectarian tribalism dressed in Christian language; and Allam warns that journalism and national-security policy have helped make Muslims a suspect category whose logic now extends to others.
Today’s applied AI sources traced the same constraint from model serving to public markets, data centers, venture strategy, and workplace agents. Stanford’s inference lecture framed the technical root: generation is sequential, often memory-bound, and increasingly defined by KV-cache movement, while the rest of the brief showed how that bottleneck is being translated into hardware valuations, powered-shell construction, policy fights, orchestration layers, and trust problems inside companies.
Harvey co-founder and chief executive Winston Weinberg argues that AI will commoditize much of the routine work product in law while increasing the value of judgment at the point where legal decisions are made. In a Knowledge Project interview with Shane Parrish, Weinberg describes how Harvey grew from a GPT-3 test on landlord-tenant questions into an $11bn legal AI company, and explains the operating discipline behind it: faster decisions, sharper prioritization, and a team built to withstand repeated failure.
John Coogan and Jordi Hays read Audemars Piguet’s Swatch “Royal Pop” as a sanctioned cheap lookalike: not a real Royal Oak substitute, but a lower rung into a brand whose entry point has moved far out of reach. Coogan also framed Cerebras’s higher IPO range and reported oversubscription as evidence that AI chip demand is being repriced around inference speed. On Trump’s China trip, he argued that tech priorities such as export controls, compute and AI access may be crowded out by Iran, oil and diplomacy.
Alex Wilhelm and Jason Calacanis treat Cerebras’s raised IPO range as a test of how much public investors will pay for future AI inference demand and the quality of contracts with customers such as OpenAI. Ori Goshen makes a parallel case that enterprise AI’s hard problem is no longer choosing one model, but routing work across models, tools and inference strategies for cost, latency and accuracy. Across OpenAI’s deployment spinout, AI21’s orchestration pitch, Magrathea Metals’ brine-based magnesium plan and OpenClaw’s fading momentum, the article frames deployment as a question of incentives, constraints and where the bottleneck actually sits.
UC San Diego professor Michael Yip argues in a Stanford Robotics Seminar that medical robotics must move beyond teleoperation if it is to address healthcare labor shortages. Current surgical robots can improve precision but still depend on a surgeon’s skill, while surgery’s scarce data, deformable tissue, safety constraints, and need for millimeter accuracy make end-to-end learning an inadequate answer on its own. Yip makes the case for a hybrid path: modern perception where it works, explicit physics and control where contact demands it, and humanoid platforms where broader hospital tasks require more general embodiment.
Stanford’s CS336 lecture on inference, taught by Percy Liang with Tatsunori Hashimoto, argues that serving language models is now a core systems problem rather than an afterthought to training. Liang’s central claim is that autoregressive Transformer generation is sequential and often memory-bound, especially because attention must repeatedly move KV-cache data rather than perform dense, easily parallelized computation. The lecture treats batching, grouped-query and latent attention, quantization, pruning, speculative decoding, continuous batching, and PagedAttention as different attempts to move fewer bytes, reuse memory better, or trade latency for throughput without degrading model quality too much.
Shrimai Prabhumoye of Mistral AI used a Stanford CS25 seminar to argue that large-language-model pretraining is becoming less a matter of adding tokens and more a question of training strategy. Drawing on studies of curriculum ordering, early reasoning data, and reinforcement as a pretraining objective, she said base models improve when they see broad data before high-quality data, encounter reasoning traces during pretraining rather than only post-training, and are rewarded for intermediate thoughts that improve prediction.
Nouamane Tazi of Hugging Face uses a Stanford CS25 seminar to argue that ultra-scale model training is less a question of adding GPUs than of managing memory, communication, batch size, and hardware topology. His central case is that 5D parallelism—data, tensor, pipeline, context, and expert parallelism—lets training runs span massive clusters only when each axis is chosen for a specific bottleneck. The practical rule, he says, is conservative: shard only as much as the workload requires, because every added parallelism dimension buys scale by spending communication, complexity, or both.
TBPN’s discussion argued that AI’s value is now being tested less in model demos than in the bottlenecks around deployment: inference speed, power, workflow integration and access to customers. Cerebras was framed as a public-market bet on faster inference, while Giga Energy’s data-center business showed how scarce powered shells have become part of the AI supply chain. The same bottleneck logic appeared outside core AI, from Audemars Piguet using Swatch as an official low-cost entry point to Augustus, with conditional OCC approval, trying to rebuild dollar clearing as a national bank.
Bloomberg Technology’s Caroline Hyde and Ed Ludlow framed Cerebras’ upsized IPO as part of a wider shift in which AI infrastructure is drawing capital across chips, data centers, power, payments and security. Bloomberg’s Rebecca Torrence said the Cerebras offering was more than 20 times oversubscribed, while other guests argued that investor demand is being supported by earnings growth, capacity constraints and expanding use cases rather than chips alone. The broadcast’s through-line was that the AI buildout is becoming a market-wide infrastructure trade, with financing, energy supply, stablecoins, cybersecurity and local hardware all pulled into the same investment case.
Ben Horowitz, co-founder of Andreessen Horowitz, uses a Stanford CS153 lecture with Anjney Midha to argue that venture capital is a systems business whose constraints keep moving. He says a16z was built in 2009 to serve entrepreneurs rather than merely allocate capital, using centralized control, small investment groups, and a deliberately constructed relationship network. In Horowitz’s account, AI has shifted the next bottlenecks toward capital, compute, electricity, policy, moats, and culture, forcing venture firms and startups to redesign around those constraints rather than rely on older software-era assumptions.
Rezolve AI chief executive Dan Wagner used a Bloomberg Technology interview to defend his hostile bid for Commerce.com as an effort to accelerate Rezolve’s push for leadership in commerce and retail AI. Wagner argued that Commerce.com’s 60,000 merchants are an underused asset held back by weak growth and limited innovation, while Rezolve’s own revenue momentum and anti-hallucination technology could make that customer base more valuable under its control.
Bloomberg’s Stacey Vanek Smith describes a $400 wager between Stanford’s Erik Brynjolfsson and Northwestern’s Robert Gordon over whether US productivity growth will average 1.8% from 2020 to 2030. Smith says recent data, including 2.9% year-over-year growth in early 2026, suggest productivity is improving, but she cautions that the figures do not show how much is due to AI. The central dispute is whether AI is making workers more productive, or whether layoffs are raising output per hour by reducing labor hours.
Circle CEO Jeremy Allaire told Bloomberg Technology that falling interest rates are not the defining constraint on the stablecoin issuer’s growth, arguing that USDC’s utility, transaction volume and network effects matter more than reserve yield alone. Pressed on Circle’s exposure to lower returns on the assets backing USDC, Allaire pointed to nearly $30 trillion of first-quarter on-chain USDC transactions, ARC’s planned launch and rising payments-network volume as evidence that Circle is trying to build a broader platform business around stablecoin activity.
Alex Kantrowitz and Ranjan Roy read Anthropic’s SpaceX compute deal as both a real answer to Claude’s capacity constraints and a piece of market theater around AI demand, financing and IPO timing. Kantrowitz argues the Colossus 1 capacity could materially ease Anthropic’s limits and sharpen its race with OpenAI; Roy cautions that explosive usage and infrastructure announcements are also serving valuation narratives. The discussion extends that frame to OpenAI trial messages, Anthropic’s Mythos security claims and AI-linked layoffs: genuine progress, they argue, is being folded into stories that remain only partly proven.
Matthias Luebken argues that coding agents such as OpenClaw are less mysterious than they appear: they are LLMs calling tools in a loop, made more useful by a runtime, shell, sessions and product hooks. In his Tavon talk, he uses Pi, a minimal coding-agent SDK, to show how that loop can be embedded inside business software, including a sales workflow where RFP emails are routed to customer-specific agent sessions and returned to users as draft replies. His architectural point is that teams should not force agents through opaque systems, but expose data, commands and controls in forms coding agents can use cleanly.
Carol Schleif, chief market strategist for Bank of Montreal, argues that the AI-driven equity rally is broader than the familiar mega-cap technology trade. In a Bloomberg Technology interview, she says earnings and revenue growth across much of the market, along with a multi-year infrastructure buildout in power, chips, materials and supply chains, are giving the rally fundamental support even as investors worry about geopolitical and energy bottlenecks.
Economist Gaurav Khanna argues that Indian high-skilled migration to the United States during the internet-era tech boom was not a simple case of brain drain. In his account, Indian programmers and computer scientists helped expand US technology production and innovation, while the prospect of migration, visa limits and return migration helped build India’s own skilled workforce and IT networks. The result, he says, was a “brain circulation” channel that contributed both to Silicon Valley’s growth and to India’s emergence as the world’s largest IT exporter.
Fryderyk Wiatrowski argues that building Viktor as an AI coworker inside Slack is not a matter of scaling a personal assistant to more users. A company-level agent gains value from shared context, shared integrations, and the ability to act where work is discussed, but those same features create harder problems around memory isolation, permissions, fragmented Slack conversations, proactivity, and tone. His case is that an “AI employee” has to be designed less like a chatbot and more like a new hire entering the company’s communication layer.