Workplace Trends

US treasury rates have been rising, albeit slowly, during 2026, but long term rates hitting new highs, the US government debt hitting $40 trillion and a new Fed chair in Kevin Warsh have all combined to make them a central player in markets today. In this post, I look at the US treasury rate movements in 2026 and try to put that movement in context not only by looking at the history of US treasury rates going back decades but also by looking at movements in government bond rates in other currencies. I then trace the effect of these rising rates on bond prices, always negative, and contrast them with equities, where the cashflows can change as rates change. The latter insight may explain the resilience of equities in 2026, even as rates have risen.

Of course you need to use open-source models if you’re an enterprise leader. Close model providers, that are now forcing data retention, are gaining immense leverage on your business if you don’t. As you connect models to your business context, they see it and learn from it, and have a track record of going after their most successful customers thanks to this information. But that’s not enough, you also need to store your data and records in open systems, or your software vendors might block you from building AI systems outside of the walled garden they have set up for you. If you can’t convince them to give you complete access to the data they manage for you, AI fortunately allows you to migrate quite fast. Once you’ve got hold of your data, you’ll need to manage how AI systems can access this data on behalf of human users, because you don’t always want Bob to see what Alice is doing in your company. That’s hard and merciless, since AI models are great at finding need-to-know errors. It takes systems that check hard access rules and models that check soft access rules. Now comes the most important part. You need to set up your own continuous training flywheel, so that you can improve your AI systems based on their interaction with your employees and your users. This is how you turn the edges of your business into AI systems your vendors and competitors cannot replicate. It’s also how you reduce deployment cost as well, as you can shrink models according to model input distribution. Those bills are getting substantial, we need to collectively become efficient if we want AI development to continue, so that matters. All of these efforts might seem daunting – they are. This is both a complete replatforming of your IT, and a complete change in the way you’re developing software, and operating your business. AI lifecycle management requires understanding human behavior and gradient descent, that’s a stretch. At Mistral, we facilitate that work by providing all primitives that you need in a single control plane, Studio, and a training platform, Forge. With our applied AI engineers and scientists working hand-in-hand with our customers, we ensure that we transfer knowledge, and that we can disappear once the systems are up and running. We deploy on our customers' infrastructure, or through our zero-data-retention hosted services, so that your edges remain your edges, and the switch button can be fully in your hand. Frontier AI can accelerate the growth of your business, but if it’s not in your hands, it’s not going to be your growth.

In late 2022 the world passed 1 TW of installed solar capacity. Just 3.5 years later we've already reached 3 TW. ➡️ The first terawatt took 68 years. ➡️ The second took 2 years. ➡️ The third took just 18 months. According to the latest Global Solar Market Outlook from SolarPower Europe, three factors have driven this acceleration: ✅ Global manufacturing capacity has expanded dramatically, ensuring supply has kept pace with soaring demand. ✅ Solar module prices have fallen around 99% since 2000, making solar the cheapest source of new electricity in much of the world. ✅ More countries are now deploying solar at scale, with annual installations measured in hundreds of gigawatts rather than tens. The result is that global solar installations have reached a scale that would have seemed unimaginable just a decade ago. The exponential growth won't continue forever and several key markets are now entering a new phase. Rapid capacity growth has started to expose system-level constraints that were largely invisible before. It is increasingly transmission bottlenecks, grid constraints, storage and system flexibility that will determine how quickly solar can continue expanding. That is increasingly where governments, utilities and developers are focusing their attention. The challenge is no longer building panels – it's building everything else around them.

According to benchmarks, a new small 3B-parameter model achieves Opus-4.5-/frontier-level coding performance. It comes with a very good technical report with lots to learn. What's fascinating is, the whole model builds on the old Qwen2.5-Coder-3B stack (yes, Qwen2.5, not Qwen3.5). So, that's a pretty clear example that highlights how much of the performance gains come from good data curation and post-training pipelines. Based on the tech report, here are some of the important pieces of their post-training stack: 1. High-signal synthetic data (math problems with credible solutions, code with tests) 2. Multiple reasoning paths for each answer 3. Filtering, filtering, filtering 4. 2-stage SFT (start with broad training, then train on hard long-reasoning samples) 5. Use target (pass@k) accuracy over validation loss for checkpoint selection 6. MGPO (MaxEnt-Guided Policy Optimization) for RLVR: basically a GRPO-style RL method with an extra weighting that favors examples that are neither too easy nor too hard for the current policy 7. Single 64k long-context RL (they found that the usual progressive context expansion hurt this model because early truncation damaged long-thinking behavior) 8. Training data order: they do Math RL, then Code RL, then STEM RL in this particular oder which they found helped overall 9. After optimizing for accuracy, they add a stage that rewards shorter correct trajectories; basically making the model more efficient without accuracy degradation 10. Offline self-distillation: they collect high-quality verified trajectories from the Math, Code, and STEM RL checkpoints, filter them, and distill them back into one unified student model 11. Instruct RL: the final stage uses rule-based validators and rubric-based reward models to improve instruction following (again, while preserving the reasoning gains) Besides, it's also really cool to see how far one can push a small 3B model, and that one can do impressive research and engineering work on a small(er) scale! They don't share the exact GPU hours for this project, but if we were to go with their previous VibeThinker 1.5B model report (which had some numbers), I'd probably say it cost around $25k to $60k. Sure, that's a lot of money, but it's not millions! (Caveat: the model is pretty new, and benchmarks could be too good to be true; need to use it in the next days to see if vibes of VibeCoder actually check out in practice. But impressive first impression! And nice post-training method write-up.)

Only 25% of wealthy families successfully preserve wealth into the second generation. Roughly 10% make it to the third generation, and just 5% sustain that wealth into the fourth. Those numbers help explain why many Family Offices are being forced to rethink their structure, priorities, and long term purpose. The traditional image of the Family Office has long been tied to scale, exclusivity, and large internal operations. Dedicated investment teams, private legal counsel, concierge services, and layered governance structures became markers of sophistication for ultra wealthy families seeking greater control over their financial lives. Now, many Family Offices are moving in a different direction. Despite continued growth in global wealth, a rising number of Family Offices are downsizing, consolidating operations, or shutting down entirely. The shift has less to do with declining wealth and more to do with rising complexity, operational costs, and changing generational priorities. Maintaining a fully staffed Family Office today requires significant expense across talent, compliance, cybersecurity, technology, and administration. For many families, especially those below the ultra large institutional level, the structure no longer delivers the efficiency it once promised. The issue is rarely investment performance alone. More often, wealth disappears because of weak governance, lack of communication, succession failures, entitlement, and growing family fragmentation over time. Generational transition is also reshaping the Family Office itself. Second and third generation family members often bring different investment philosophies, levels of involvement, and long term priorities. As families spread across multiple regions and jurisdictions, alignment becomes more difficult and governance grows more complicated. In response, many families are adopting leaner structures focused on oversight and strategy while outsourcing specialized functions to external partners. Investment management, estate planning, reporting, cybersecurity, and administrative services can now be handled externally with institutional quality support. Technology has accelerated this shift, allowing smaller teams to operate with greater efficiency and visibility than ever before. The conversation is also becoming more intentional. Many families are no longer measuring success by the size of their operation. Instead, the focus has shifted toward governance, communication, succession planning, and long term family cohesion. In many cases, a smaller and more focused Family Office structure may be better suited for preserving wealth across generations than a large internal organization weighed down by complexity. The Family Office industry is still growing globally, but the model itself is changing. The future Family Office will likely be defined less by size and more by adaptability, clarity, and strategic coordination.

From artificial intelligence reshaping job roles to hiring booms beyond metros, the world of work is changing fast. As India's Class of 2026 prepares to enter the workforce, two questions are top of mind: where are the real opportunities, and what does it take to land them? LinkedIn’s Grads’ Guide 2026 tracked the fastest-growing opportunities for career starters across India — by job title, industry, function, and region. Here’s what stood out. AI Specialist, Generative AI Engineer, and Digital Content Creator emerge as the top three fastest-growing job titles, indicating how the rise of AI and digital platforms is opening up entirely new career pathways. Meanwhile, Human Resources (HR) and Consulting lead the fastest-growing functions for graduates. This rise in demand for fresh talent in HR “is being driven by two forces — a genuine shortage of top talent, making hiring increasingly competitive, and the growing need to manage large-scale workforce changes,” says Arjun Prakash, Founder, Pivot. In consulting, the hiring momentum is being driven by several structural shifts — including AI and cloud transformation, GCC expansion, strong global demand for Indian tech talent, and maturing start-ups increasingly turning to first-time consulting projects to drive profitability, notes Arjun. “The rise of specialist firms across HR, finance, and operations consulting is further accelerating this trend,” he adds. Meanwhile, Utilities and Education are the top two fastest-growing industries actively hiring graduates, shows LinkedIn data. Regional hiring growth patterns further reinforce how opportunities for early talent are expanding beyond traditional hubs. Vijayawada occupies the top spot amongst the fastest-growing regions for early-career hiring, followed by Kolkata and Bhopal. “Better cost efficiencies in tier-2 and tier-3 cities, lower attrition driven by proximity to home, and the adoption of distributed work models are driving this shift. At the same time, improving digital infrastructure and broader access to talent make it easier for companies to evaluate talent beyond metros,” notes Vaibhav Gadodia, CTO, Nagarro. For the Class of 2026, the opportunity is real, but it doesn’t always look the way you expect it to. ➡️ Swipe through the carousel for the full data breakdown. How can freshers make their move in India’s new world of work? Share your thoughts using #GradsGuide2026 Editor: Nakul Ghai Data insights: Alejandra Budar, Caroline Liongosari, and LinkedIn's Economic Graph Graphics: Arunagiri Ramadurai

Our research center in Princeton has become a magnet for healthcare AI expertise. Every time I catch up with Dorin Comaniciu and the team there, conversations quickly move from what’s possible to what really matters in healthcare delivery. Take for instance, our work on what we call the Operational Twin, an advisory service. It starts with creating a virtual representation of a clinical department, reflecting how patients, staff, and equipment interact in everyday operations so that different scenarios can be explored more safely and at scale. By simulating billions of scenarios representing dynamic conditions, AI agents learn how operational decisions shape outcomes. They can begin to anticipate bottlenecks and understand the long-term impact of short-term choices. The goal is more efficient planning of patient schedules, staffing, and equipment use, aligning daily decisions with broader clinical and organizational priorities. This becomes even more relevant as clinical innovations accelerate workflows. Faster scanning technologies such as Deep Resolve can shorten patient timeslots and an Operational Twin can help organizations adapt by optimizing schedules and resources to fully realize gains in speed and throughput. At its core, this work is about creating clarity in complex systems so that action becomes more precise and more purposeful. We see a similar principle in clinical innovation. With photon counting CT, we can visualize the heart in extraordinary detail, including structures inside the left ventricle that were previously difficult to see clearly. That deeper insight is captured by a Foundation Model that could help physicians guide ablation therapies with greater precision and confidence, especially when combined with live ultrasound to support real-time decision making in the procedure room. In both cases, whether in clinical imaging or in operations, the ambition is the same: better insight leading to better decisions at the moments that matter most for patients. 𝘋𝘪𝘴𝘤𝘭𝘢𝘪𝘮𝘦𝘳: 𝘛𝘩𝘦 𝘱𝘳𝘰𝘥𝘶𝘤𝘵𝘴/𝘧𝘦𝘢𝘵𝘶𝘳𝘦𝘴 𝘢𝘯𝘥/𝘰𝘳 𝘴𝘦𝘳𝘷𝘪𝘤𝘦 𝘰𝘧𝘧𝘦𝘳𝘪𝘯𝘨𝘴 𝘮𝘦𝘯𝘵𝘪𝘰𝘯𝘦𝘥 𝘩𝘦𝘳𝘦 𝘢𝘳𝘦 𝘯𝘰𝘵 𝘺𝘦𝘵 𝘢𝘷𝘢𝘪𝘭𝘢𝘣𝘭𝘦 𝘪𝘯 𝘢𝘭𝘭 𝘤𝘰𝘶𝘯𝘵𝘳𝘪𝘦𝘴. 𝘐𝘧 𝘵𝘩𝘦𝘴𝘦 𝘴𝘦𝘳𝘷𝘪𝘤𝘦𝘴 𝘢𝘳𝘦 𝘯𝘰𝘵 𝘮𝘢𝘳𝘬𝘦𝘵𝘦𝘥 𝘪𝘯 𝘤𝘦𝘳𝘵𝘢𝘪𝘯 𝘤𝘰𝘶𝘯𝘵𝘳𝘪𝘦𝘴 𝘧𝘰𝘳 𝘭𝘦𝘨𝘢𝘭 𝘰𝘳 𝘰𝘵𝘩𝘦𝘳 𝘳𝘦𝘢𝘴𝘰𝘯𝘴, 𝘵𝘩𝘦 𝘴𝘦𝘳𝘷𝘪𝘤𝘦 𝘰𝘧𝘧𝘦𝘳𝘪𝘯𝘨𝘴 𝘤𝘢𝘯𝘯𝘰𝘵 𝘣𝘦 𝘨𝘶𝘢𝘳𝘢𝘯𝘵𝘦𝘦𝘥. 𝘍𝘰𝘳 𝘮𝘰𝘳𝘦 𝘪𝘯𝘧𝘰𝘳𝘮𝘢𝘵𝘪𝘰𝘯, 𝘱𝘭𝘦𝘢𝘴𝘦 𝘤𝘰𝘯𝘵𝘢𝘤𝘵 𝘺𝘰𝘶𝘳 𝘭𝘰𝘤𝘢𝘭 𝘚𝘪𝘦𝘮𝘦𝘯𝘴 𝘏𝘦𝘢𝘭𝘵𝘩𝘪𝘯𝘦𝘦𝘳𝘴 𝘳𝘦𝘱𝘳𝘦𝘴𝘦𝘯𝘵𝘢𝘵𝘪𝘷𝘦.

Performance Management in the Age of AI: the new 3‑Dimensional Model For decades, the 9‑box grid shaped how organizations assessed talent—mapping individuals along two familiar axes: ✔ Business performance (“what”) ✔ Behaviors or potential (“how”) Over time, many companies moved away from this model, concluding it oversimplified the complexity of human performance and sometimes reinforced bias more than it reduced it. AI is fundamentally reshaping work, shortening the lifecycle of skills and creating new capability demands at a pace conventional frameworks were never designed to keep up with. As a result, a new paradigm for performance management is emerging. Organizations are starting to consider a three‑dimensional approach to performance—one that integrates not just what people deliver and how they behave, but also how they grow. The new 3D model consists of three axis: 1. Business Results: Measures impact, delivery, and contribution to outcomes. 2. Behaviors / Ways of Working: Captures collaboration, leadership etc. and.. 3. Skills Development: Assesses capability building, learning velocity, and readiness for future roles. The third axis reflects a simple reality: In an AI‑driven workforce, continuous skills development is no longer optional—it’s strategic. IBM has begun to formalize this multidimensional view in its talent and rewards model. Their approach includes: 1. Integrating skills into pay: Base pay and equity linked to skill progression. 2. Balancing objectives: Business and skills goals carry equal weight 3. Future skills visibility: Regular communication on evolving skill requirements see: https://lnkd.in/eTDE-XmE Not every organization can replicate this model at scale, but it illustrates where performance management is heading. The central questions are shifting. Not just: “Did someone deliver results?” But also: “Are they developing the skills the organization will need next?” and “Are they learning at the speed the environment requires?” The move from a 2D grid to a 3D, capability‑driven framework may become one of the most consequential shifts in performance management in the age of AI—signaling a future where growth, adaptability, and skill relevance stand on equal footing with results.

Notional defaults (red) remain low, even as bankruptcy filings (yellow) have picked up. ➡️ This is not a classical credit cycle. Bond default rates dollar-weighted have stayed MUTED, far below prior stress episodes, reflecting ample liquidity, refinancing capacity, and strong issuer differentiation. ➡️ Bankruptcy filing numbers in contrast have RISEN, concentrated among smaller, weaker firms, reflecting the K-shaped dynamic seen across the economy, consumers, and equity margin trends, and pointing to idiosyncratic stress rather than systemic deterioration, in our view. ➡️ In our 2026 Global Outlook, we write about how an environment of greater DISPERSION makes manager selection, due diligence, workout capabilities and track records even more crucial. It gives established lenders with strong documentation and proven expertise an edge.

NEW ANALYSIS: Electric vehicles are entering the mid-transition space starting to replace ICE vehicles in more and more markets. The transition is already underway. Global EV numbers have grown from 1.2 million in 2015 to nearly 60 million today. History shows that shifts like this can happen faster than expected: in the early 20th-century US, horses and mules virtually vanished from roads in under 30 years. As with the rise of the car, today’s transition is shaped as much by policy and politics as by technology. ICE vehicles didn’t dominate through technical superiority alone—they were supported by massive public investment in roads, urban design, and highways funded by fuel taxes. EVs are well placed to move even faster. They directly replace ICE vehicles while being cleaner, cheaper, and quieter to operate. And past transitions suggest that like-for-like replacements—think black-and-white to colour TV—tend to spread far more quickly than entirely new products. Our new report by the Centre for Net Zero (Octopus Energy Group)'s excellent Andy Hackett, Izzy Woolgar, RMI's Yuki Numata and Laurens Speelman and me at Environmental Change Institute (ECI), University of Oxford describe how EVs are posed to enter a next phase in it's adoption curve. This is the phase of 'system integration', where integration of EVs into the broader energy and transport system (think vehicle to grid, flexible charging, widespread and equitable charging, battery recycling) becomes more and more important, alongside reducing costs, intense competition, increasing quality and efficiency, and increasing supporting technologies. This new phase represents new opportunities and new challenges both for policy makers and business which we unpack in this report. You can read the report here: https://lnkd.in/eRNdpMj6