AI Investment Insights

AI is entering its cloud-cost moment. Uber reportedly burned through its AI coding tools budget far faster than expected. GitHub Copilot is moving further toward usage-based economics. Across the market, the economics of AI experimentation are becoming harder to ignore. For much of the past year, many companies treated AI usage as if it were effectively unlimited. That was never going to last. By late spring, the assumption started to visibly break. During the first phase of enterprise AI adoption, the labs absorbed much of the economics while companies experimented. Heavy users often extracted far more value than they paid for, and that made rapid experimentation feel almost frictionless. But once AI spend becomes a real line item in the P&L, behavior changes. A few things are becoming clear. The bills are arriving. Teams that left the throttle wide open are now seeing usage numbers they never planned for. What looked like harmless experimentation is suddenly under review. Pricing is shifting underneath the market. Flat per-seat models are giving way to consumption-based billing, turning predictable software spend into something more variable and harder to forecast. Compute is becoming a real constraint. Not just GPUs, but power, cooling, and physical capacity. None of that gets solved on a quarterly timeline. Capital is moving down the stack. Inference, hosting, and power now matter as much as the models themselves. The model is no longer the whole prize. New releases are still impressive, and capability still matters. But enterprise value is increasingly determined by everything around the model: how it is run, governed, and scaled. One of the clearest signals is that even AI labs are expanding their deployment, enterprise, and services teams. The challenge is no longer only building intelligence. It is making that intelligence work inside real enterprises, with real budgets, real governance, and real consequences when systems fail. If the conversation is moving from capability to economics, three questions now matter: Do you understand the true cost of your AI workloads in production? Do you have clear ownership and accountability for every business-critical AI system? And do you know which workloads require sovereign, private, or regulated environments before cost, compliance, or capacity force the decision for you? Those questions may matter more over the next 12 months than which model you choose. AI was never only a model problem. It is an operating problem. When AI was an experiment, success was defined by what you could demo. In production, it is defined by what you can run. The winners of this next phase will not be the companies that spent the most or ran the flashiest pilots. They will be the ones who already knew how to operate AI while everyone else was still admiring the model.

The AI Investment Boom: From Compute Gold Rush to Strategic Moats After sharing my perspective at the 1st Azerbaijan Investment Forum, I’m equally struck by how capital allocation in AI is reshaping the landscape in 2025. Investment trends this year are telling a powerful story, one that goes beyond hype: Capital Flows Are Redefining the AI Map ✔ Enterprise AI Spend surged 6×, from $2.3B in 2023 to $13.8B in 2024. Pilots turned into production, and AI is now embedded in P&L, not just PowerPoints. ✔ DeepSeek’s $6M Breakthrough showed that frontier-class models don’t always need 9-figure training budgets. This signals a new investment thesis: efficiency and ingenuity can compete with brute force. ✔ $1.5B Anthropic Settlement reminded investors that data rights and compliance aren’t “nice-to-haves” they’re billion-dollar risks. The smart money is flowing into AI compliance, security, and eval tooling. ✅The investment signals show us four durable moats: Data & Distribution: The moat isn’t just the model; it’s who controls data pipelines and user distribution. Infrastructure: Nvidia, AMD, TSMC, and hyperscalers still dominate, but efficiency challengers are emerging. Ecosystem Building: Talent + infrastructure = defensible advantage. The players creating robust agent ecosystems will own the platform layer. ✅Signals Investors Can’t Ignore: Thin Wrappers Still Work – Despite the hype cycle, lightweight apps built on top of foundation models keep winning niches. Even frontier labs are now funding or acquiring wrappers because distribution matters. Top Models Aren’t Everything – GPT-5 and Gemini dominate benchmarks, but they don’t dominate every use case. Open-source proved that agility and community adoption can be just as powerful. Talent Concentration Is Destiny – The biggest AI clusters (Bay Area, Beijing, Abu Dhabi) attract talent and capital, reinforcing themselves. Smart investors back ecosystems, not just single models. The Implication for Investors 👉 Not every “AI startup” will survive. Many will be commoditized. 👉 But data, distribution, compliance, and ecosystems are proving to be the real durable plays. 👉 The global map is also shifting: China (DeepSeek) is a credible challenger, while UAE and Singapore are positioning as investor-friendly AI hubs. Grateful to the hosts, ministers, and fellow panelists for the opportunity to exchange ideas together. The energy and insights from this group were truly inspiring: Samir Sharifov – Deputy Prime Minister Mikayil Jabbarov – Minister of Economy of the Republic of Azerbaijan simonida kordic – Minister of Tourism of Montenegro Jagoda Lazarevicć – Minister of Domestic and Foreign Trade, Serbia Nouriel Roubini, MAHDI ALADEL, Lisa Bodell and Fariz JAFAROV (Executive Director, 4SİM Azərbaycan / C4IR Azerbaijan). As I emphasized in Baku: this is the slowest AI will ever be. For investors and innovators alike, the time to place smart bets is now.

Markets aren't always rational, particularly in the short term, but market reactions to last week’s earnings announcements from some of the most scrutinized companies on earth — Meta, Google and Microsoft — caught my attention as an important signal. My interpretation is that while all three companies are pouring billions into AI-related capex, Wall Street is increasingly skeptical about whether consumer-facing AI (like Meta’s “personal superintelligence”) can justify the massive capex and deliver sufficient TAM. Meanwhile, Google and Microsoft are being given much more license to invest ahead of revenue and build capacity to meet existing and projected demand for enterprise applications, even if the ROI isn’t yet fully visible. What strikes me is how AI investment is mirroring to some extent the “growth at all costs” playbook — but with capacity spending. Meta's decline suggests to me that investor confidence is wearing thin for consumer-facing AI, while the market seems to be rewarding enterprise software that creates business value with AI. And that seems rational for the longer term given how enterprise software that incorporates AI can transform end-to-end business systems for customers.

NVIDIA reported earnings yesterday, and, as is tradition, they crushed expectations, guided conservatively, and the stock promptly fell 3% because when you’re priced for perfection, even dominance is a mild disappointment. But let’s ignore the stock market tantrum for a moment and parse Jensen Huang's earnings call commentary for industry context: 🚀 AI Demand is Still in Hyper-Growth Mode. Data Center revenue surged to $35.6B (up 93% YoY). Blackwell is NVIDIA's fastest-ramping product ever—$11B in its first full quarter, not even a year after it was first announced. Jensen notes "It will be common for Blackwell clusters to start with 100,000 GPUs". 🧴 Inference is the Bottleneck. Reasoning models like OpenAI's GPT-4.5, DeepSeek AI-R1, and Grok-3 require 100x more compute per query than their early ancestors. AI is moving beyond one-shot inference to multi-step reasoning, chain-of-thought prompting, and autonomous agent workflows. Blackwell was designed for this shift, delivering 25x higher token throughput and 20x lower cost vs. Hopper. 📈 3 Scaling Laws. Jensen identified three major AI scaling trends that are accelerating demand for AI infrastructure: (1) Pretraining scaling (more data, larger models) (2) Post-training scaling (fine-tuning, reinforcement learning) (3) Inference-time scaling (longer reasoning chains, chain-of-thought AI, more synthetic data generation). 💰 Who's Buying? Cloud Service Providers (CSPs) still make up about 50% of NVIDIA's Data Center revenue, and their demand nearly doubled YoY but many enterprises are also investing in their own AI compute instead of relying solely on cloud providers 🍟 Custom Silicon and the ASIC vs. GPU Debate. Big Tech is building custom AI ASICs (Google has TPUs, Amazon has Trainium, Inferentia) to reduce dependency on NVIDIA but Jensen dismissed the notion that custom silicon would challenge NVIDIA’s dominance. GPUs remain more flexible across training, inference, and different AI models, while ASICs are often limited in their use cases. He flagged the CUDA ecosystem as a major competitive moat. 🛰️ The Next Frontier. Jensen repeatedly emphasized “agentic AI” and “physical AI” as the next major trends. The first AI boom was digital—models that generate text, images, and video. The next phase is AI that acts and interacts with the physical world. The market may worry about Nvidia's forward guidance but its hard to discount a company that controls everything from the chips to the networking (NVLink, InfiniBand), software (CUDA, TensorRT) and system-level AI solutions.

Investors can’t get enough of AI companies but the signs of overvaluation in AI are flashing red. The AI sector’s near-euphoric investment surge is showing clear signs of extreme overvaluation and the data is striking: 1️⃣ $73 billion in global VC funding flowed into AI startups in Q1 2025 which is nearly 60% of all venture deals. 2️⃣ Public AI equities have jumped 46%, capturing one-third of $46 trillion in global market-cap gains over five years. 3️⃣ Many startups now trade at 20x–50x revenue multiples, far above traditional tech benchmarks. 𝗧𝗵𝗲 𝗥𝗶𝘀𝗲 𝗼𝗳 “𝗖𝗶𝗿𝗰𝘂𝗹𝗮𝗿 𝗜𝗻𝘃𝗲𝘀𝘁𝗶𝗻𝗴” 𝗶𝗻 𝗔𝗜 A new trend called circular investing is emerging where AI companies fund, supply, and buy from one another, creating financial feedback loops reminiscent of the dot-com bubble. 𝗥𝗲𝗰𝗲𝗻𝘁 𝗲𝘅𝗮𝗺𝗽𝗹𝗲𝘀: Nvidia ↔ OpenAI: Nvidia’s pledged $100 B investment mirrored by OpenAI’s multibillion-dollar chip orders. AMD ↔ OpenAI: Warrants grant OpenAI equity in exchange for purchase commitments. Oracle ↔ OpenAI: A $300 B compute deal tied to Nvidia’s investment cycle. 𝗩𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 𝗪𝗮𝗿𝗻𝗶𝗻𝗴𝘀 OpenAI’s reported $1 trillion valuation stretches investor imagination on future earnings. Analysts and institutions from Wall Street to the IMF warn that current valuations assume flawless execution with almost no margin for error. 𝗔 𝗕𝗮𝗹𝗮𝗻𝗰𝗲𝗱 𝗩𝗶𝗲𝘄 Mark A. Jamison (AEI) in Barrons argues the picture isn’t purely alarming: • AI’s profits remain concentrated in cash-rich incumbents with genuine innovation. • Circular deals can represent strategic risk-sharing, not mania. • Unlike the dot-com era, much of this boom is funded by free cash flow, not debt. Yet, the sector still faces headwinds like soaring energy needs, regulatory scrutiny, and uneven enterprise adoption. 𝗧𝗵𝗲 𝗧𝗮𝗸𝗲𝗮𝘄𝗮𝘆 AI promises a multi-decade technological supercycle, but we must be mindful of the risks inherent in overvaluation and circular financial structures. Prudent investment discipline, transparency, and realistic expectations will be crucial to avoiding bubble-like pitfalls and ensuring lasting value creation.

AI Opportunity Map based on Selected Use Cases along the Value Chain in Banking and Payments💡 The transformative potential of AI unfolds in a dual capacity: Firstly, increasing revenue growth in customer-centric areas such as product management, client engagement, and customer relationship management. Concurrently, it drives higher efficiency in underlying processes, including operations & execution, as well as risk & compliance. According to the European Investment Bank (EIB) (2021), AI holds considerable promise in the banking sector: “In banking, it has been estimated that AI could increase banks’ revenues by as much as 30 % and potentially reduce their costs by 25 % or more.” This highlights AI’s strategic value in the payments sector, outlining a path toward customer-focused growth and cost-efficient operations. The adoption of AI across these dimensions signals a future marked by innovation, operational efficiency, and enduring success 🤖 Drawing from the opportunity areas depicted in the AI opportunity map, three key potentials for financial services institutions emerge: Potential 1️⃣: Tailored growth AI empowers organizations to create highly personalized solutions through the utilization of data-driven insights. Customized solutions not only boost customer loyalty but also sharpen a company’s competitive advantage, leading to higher revenues. Furthermore, AI-driven strategies in customer engagement refine personal relationship management, achieving this with impressive cost-efficiency. Potential 2️⃣: Streamlined operations and cost efficiency AI offers advantages that go beyond enhancing customer interactions, significantly improving operational efficiency on the back end. By incorporating automation and refining processes, AI leads to increased efficiency and substantial cost savings. This allows for the strategic reallocation of resources towards innovation and other critical investments, thereby creating a significant competitive advantage. Potential 3️⃣: Sophisticated Risk Management and Regulatory Compliance AI is increasingly recognized as an essential tool for real-time risk detection and adherence to regulatory compliance. It allows organizations to proactively identify and mitigate risks, thus ensuring regulatory compliance and protecting both financial stability and reputation. Integrating artificial intelligence (AI) into banking and payment systems has become a strategic imperative. This is due to the swift advancement of AI technology and the fact that numerous entities have already begun adopting this innovative technology. To create an “AI playground”, banks must address new risks, ensure robust infrastructure, and foster a culture of playing around as a basis for future differentiation 🚀 Source: Arkwright Consulting - https://bit.ly/3wwBJLb #Innovation #Fintech #Banking #FinancialServices #Payments #Lending #KYC #AML #AI #MachineLearning #Data #Cloud #Automation #LLMs #GenAI 

There are two kinds of AI strategies right now: AI for headline value, and AI for production value. Most of what’s out there is theater. Bolting a chatbot using RAG onto a broken process and calling it transformation. Sharing an obscure demo and claiming industry-breaking innovation. The receipts are coming. When the hype settles and boards start asking what AI actually produced, a lot of companies won’t have an answer. Many of the companies that think they understand AI are about to discover they don’t. Because AI is not a website feature. It is not a demo. It is not a slide in an investor presentation. It is a new production system. The companies seeing real results have moved AI deep into the machinery of their business. If AI can’t access the tools your company runs on, it can’t meaningfully improve outcomes for your clients. Most companies are still feeding one client at a time into an off-the-shelf LLM. The winners are training propensity models on decades of proprietary data, supercharging them with LLMs, and delivering through the last mile via chat, text, voice, and traditional interfaces. Historical data is the engine. LLMs are the delivery mechanism. The highest-impact use cases I’m seeing pair AI with deterministic systems. AI by itself is often unpredictable. AI executing against defined rules, thresholds, and workflows is precise. For us, that combination is driving materially higher close rates and allowing AI agents in production to execute hours of repetitive work every day. On recent earnings calls, Visa said its AI-powered risk tools helped prevent more than $10 billion in fraud. Lowe's Companies, Inc.’s said its AI-enabled quoting tool reduced quote generation from days to minutes. Citi said AI-driven code reviews created roughly 100,000 hours of weekly capacity. Many companies are asking their teams: “Are you using AI?” They should be asking: “What did AI produce?” That’s the difference between headline value and production value. And over the next few years, that difference is going to get very expensive.

#AI disruption sent shockwaves through the $3.5 trillion private credit market this week, as investors in some of the largest private credit funds asked for their money back and were told no. Stay with me for two minutes to understand what happened - and how this might affect your 401(k): Private credit rarely makes headlines. The sector emerged after 2008, when banks pulled back from lending to mid-sized companies, and new funds filled that gap by raising capital and lending it to private businesses. The appeal was simple: higher returns than traditional bonds, but your capital is harder to withdraw. For the last five years, the steadiest revenue in these portfolios has come from software companies. If your company pays for Salesforce or Workday, you probably renew every year because you've built your operations around it, and switching is expensive. That recurring, hard-to-cancel revenue is ideal collateral, and software grew to 29% of these loan portfolios. AI is now eroding the collateral those loans were built on. Enterprise software stocks are down 25-30% from their highs as AI reduces the headcount that needs licenses and makes it possible for companies to build internal tools that replace off-the-shelf software. That is AI disruption showing up not as a headline about the future, but as a balance sheet problem today. Investors realized this and sought their money back, with mixed results reported widely: Cliffwater's $33 billion fund capped withdrawals at 7% after investors requested 14%. Blackstone, BlackRock, and Blue Owl all faced record requests this quarter, and several froze redemptions entirely. I lead a large global foundation and oversee billions in endowments as a board director on investment committees and I've also spent a decade studying how AI changes economic structures. In recent years, those worlds have converged. The question of what happens to software-backed lending when AI erodes the revenue underneath has been building. UBS estimates that 25-35% of these portfolios face elevated AI disruption risk. This quarter, the convergence became visible: redemption caps, loan markdowns, and fund managers telling investors no. Until last year, private credit was restricted to institutional investors and the wealthy. Then, Executive Order 14330 opened 401(k) plans to the asset class for the first time, meaning 90 million Americans can now enter a market that institutional investors are trying to exit. Retirement savings are increasingly tied to financial vehicles where capital moves faster than our ability to measure what AI is doing to the assets inside them. Everyone is debating how AI will reshape products and jobs. The debt markets are ahead of that conversation, repricing the entire premise of enterprise software. This might be the first clear signal of bigger disruptions to come. David Ramage Sophia Tsai Karen Gill John A. Barker The Patrick J. McGovern Foundation

“If it’s not AI, I don’t want it” – a VC headed to Monaco for summer Q2'25 data* shows AI companies are securing significantly larger rounds across sectors, with median deal sizes hitting $4.6M – over $1M above the broader market. In Q2’25, the AI premium was strongest in Auto Tech which saw AI companies securing deals $20.6M larger than traditional peers (lead by Applied Intuition's $600M Series F at $15B valuation), followed by Robotics and Cybersecurity with median deal premiums of $10.7M and $6.4M respectively. The AI premium extends beyond funding to company performance and trajectory metrics. AI companies consistently score higher on our Mosaic Score (success probability) and Commercial Maturity (ability to compete and partner) metrics, proving their fundamentals justify investor confidence. Why are AI companies commanding these premiums? 1) Capital-intensive development cycles AI companies often require dramatically more upfront investment for compute infrastructure, data acquisition, and model training before achieving product-market fit, necessitating larger initial rounds to reach meaningful milestones. 2) Longer runway to defensibility Unlike traditional SaaS where competitive advantages emerge quickly, AI companies need 12-18 months of continuous model refinement and data collection to build meaningful moats, requiring sustained funding through extended R&D phases. 3) Premium for hybrid expertise The most successful AI companies combine rare AI/ML talent with deep domain expertise (like automotive engineers for autonomous driving), creating interdisciplinary teams that command higher compensation. 4) Infrastructure-first business models AI companies often build foundational platforms (like simulation environments or data processing pipelines) that require significant upfront investment but can later support multiple product lines and customer segments. The AI premium continues to reflect investors' "go big or go home" approach; making concentrated bets on AI teams they believe can capture outsized market share. The AI premium signals more than just funding enthusiasm – it's recognition that AI-first companies are simultaneously disrupting the last two decades of companies and building the infrastructure for tomorrow's economy. *Data from CB Insights’ State of Venture Q2’25 report. Explore the latest data on what happened last quarter across the startup ecosystem at the link in the comments.

💡 Why Blackstone’s recent investment is important for you? Blackstone has made a groundbreaking $300M investment in DDN, valuing the California-based AI data company at $5B. This move highlights the private equity giant’s growing focus on artificial intelligence and its supporting infrastructure. 📊 Key Highlights: 📌 DDN's Role in AI Growth: DDN provides tools to manage and analyze massive datasets, a crucial element in AI model training and deployment. They power some of the largest AI projects, including Elon Musk’s Colossus supercomputer. 📌 Blackstone’s Broader AI Push: Beyond DDN, Blackstone has invested heavily in data centers and chip-supporting companies, including a $7B deal with Digital Realty and a $16B acquisition of Asian data center operator AirTrunk. 📌 Strategic Rationale: With the explosion of AI applications, efficient data handling is non-negotiable. DDN’s solutions aim to make AI deployments more cost-effective and scalable, positioning it as a leader in the AI ecosystem. 📈 What This Means for the Broader Market: 📌 AI Infrastructure is Booming: From data centers to AI chips, the backbone of AI growth is becoming an attractive investment theme. IPO Potential: DDN’s growth trajectory suggests it could go public soon, offering new opportunities for investors. 📌 Sector Evolution: This underscores a shift toward strategic investments in AI-enabling technologies, highlighting the symbiotic relationship between private equity and tech innovation. 💡 Investor Takeaways: 📌Diversify into AI Infrastructure: AI isn’t just about software—consider the enabling technologies like data centers and hardware. 📌Long-Term Growth Opportunity: With AI adoption accelerating, companies that support its infrastructure are poised for exponential growth. 📌Stay Informed: The competitive landscape is heating up, making it crucial to follow developments in the AI ecosystem. 💬 Are you positioning your portfolio to capture the AI wave? #AI