AI is no longer just helping people research stocks. Some AI agents are now actually trading them. Robinhood says more than 150,000 customers have opened agentic trading accounts since the service launched in May, and AI agents are using Robinhood’s tools almost 30 million times a day.  So what are these AI traders buying? In one test, AI agents from Claude, Gemini and ChatGPT were each given $500 and asked to trade stocks and ETFs. The results were very different. Claude made 2.9%, Gemini gained 0.9%, while ChatGPT lost 0.7% over the three-day test. Claude and ChatGPT mainly followed momentum strategies, while Gemini focused heavily on Nvidia.  That part really caught my attention. AI agents don’t all “think” the same way. Give them the same money and the same market, and they can still

Still working towards mine! 😅 This year has had its ups and downs, with some good calls, a few mistakes, and plenty of lessons along the way. I’ve learned that investing is a long game, and not every move needs to be perfect. There’s still time left in 2026, so I’m keeping focused on my goals and making the most of the opportunities ahead. For anyone still chasing their goals too — keep going! Every step forward counts, even when it doesn’t feel like it. We’ve still got time to finish the year strong. 💪📈

$Amazon.com(AMZN)$  is looking at a different way to pay for its huge AI spending. Reports say Amazon is considering moving about $8 billion of Nvidia AI chips into a special company called an SPV. Outside investors would help fund the chips. Amazon would then lease the chips back and continue using them for AWS and AI. The interesting part? The outside investors could receive up to a 10% equity stake in the SPV, meaning they would have an ownership interest in the company holding the chips. Why does this matter? AI chips are extremely expensive, and Amazon is spending billions on data centres and computing power. Instead of Amazon paying for all the chips itself, outside investors could help fund them. This could reduce the amount of mo

Everyone is watching Nvidia and the big AI data centres. I’m looking at a different part of the semiconductor chain. $ON Semiconductor(ON)$ Semiconductor makes chips focused on power management, sensing and control — important technologies for cars, industrial equipment, automation and AI infrastructure. The big catalyst right now is its planned acquisition of Synaptics. So what does Synaptics actually do? Synaptics develops chips and technology that help connected devices see, hear, connect and interact. Its products cover touch and display, biometrics, wireless connectivity, audio, video, vision and security processing. It also has Astra, an AI-native platform designed to bring AI processing directly into devices rather than relying entirely

B — Hyperscaler AI capex. The biggest question for me is whether spending keeps accelerating or starts getting more selective. That could have a major ripple effect across GPUs, networking, data centres and the wider AI supply chain.

I’d pick D — whether AI can actually justify the spending. The investment in chips, data centres and infrastructure is enormous, so eventually the numbers have to catch up with the narrative. Revenue growth, margins and actual returns on that spending will tell us whether the AI boom is creating durable profits or simply requiring bigger and bigger investment.

If I had $100,000 in SRS with a 10+ year horizon, I wouldn’t want it sitting entirely in cash. I’d be more comfortable with a diversified mix, using broad ETFs as the core and adding some individual stocks or REITs for different sources of growth and income. With that kind of timeframe, I’d be more focused on compounding and diversification than trying to time the market. The key for me would be making sure the risk level matches the long-term goal rather than chasing the highest possible return.

For me, earnings are the one to watch in October. After so much focus on expectations, I want to see whether companies can actually deliver the growth investors are pricing in. Guidance will probably be just as important as the headline numbers.

This Broadcom–Anthropic deal caught my attention because it goes much further than a normal chip-supply agreement. $Broadcom(AVGO)$   has agreed to provide Anthropic with up to $42 billion in financing to help fund its infrastructure spending, according to Anthropic’s IPO filing. At the same time, Broadcom is involved in supplying the hardware and leasing equipment to Anthropic. And there is another important piece: Anthropic is expected to become Broadcom’s largest custom-chip customer in 2027. So Broadcom isn’t simply selling the picks and shovels for the AI buildout. It’s potentially helping finance the customer buying those picks and shovels. That’s a fascinating structure. Anthropic has committed to $125.2 billion of TPU comput

$Accenture PLC(ACN)$  is my Stock of the Day to watch after a surprisingly strong earnings reaction. The interesting part isn’t simply that earnings beat expectations. It’s what the results say about the debate around AI and the future of consulting and technology services. For a while, one of the big questions around companies like Accenture has been pretty straightforward: If AI can automate more work, won’t businesses eventually need fewer consultants and technology professionals? Yesterday’s results offered a different data point. Accenture reported Q4 revenue of $18.68 billion, above expectations, while adjusted EPS came in at $3.29. The company also reported $22.17 billion in quarterly bookings and a record $84.5 billion of bookings

When I look at a company, one thing I find interesting is not just how much it sells. It’s how difficult it would be for customers to leave. Some businesses have products people can replace in five minutes. Others become part of a customer’s daily routine, business operations or entire ecosystem. That difference can be enormous. Think about the software a company uses to run payroll, manage accounting, store data or communicate with customers. Once thousands of employees are trained on a system, years of information are stored there and other systems are connected to it, switching isn’t as simple as cancelling a subscription. There is a cost to leaving. There is a learning curve. There is disruption. And sometimes there is simply no good reason for a customer to take the risk.

Everyone is watching Nvidia. Then came the memory trade. Now another part of the AI infrastructure chain is starting to get attention: Networking. $Celestica(CLS)$   — sits in a part of the market that doesn’t get nearly as much attention as GPUs or HBM, but increasingly matters as AI data centres scale. The problem is simple: More AI = more data moving between servers. And eventually, computing power isn’t the only bottleneck. Bandwidth becomes the bottleneck. Fresh analyst coverage is highlighting this exact theme, with networking and optical hardware increasingly viewed as critical infrastructure for next-generation AI systems. Celestica is also being linked to programs involving Alphabet’s custom TPUs, OpenAI’s next-generation ch

Micron didn’t just beat expectations. It obliterated its own previous quarter — and then guided even higher. Q4 revenue hit $54.23B, up from $41.46B in Q3 and $11.32B a year ago. Non-GAAP EPS came in at $33.42, while gross margin reached 87%.  But here’s the part I find more interesting: 💰 Customers are locking in supply Micron’s long-term strategic customer agreements have jumped from $22B in June to $32B. That’s a big change for a business historically known for brutal boom-and-bust memory cycles. Reuters reports Micron has already secured agreements covering most of its expected 2027 output.  In other words, this isn’t simply: “AI demand is strong.” It’s increasingly: “Customers are willing to commit billions to make sure they actually get the memory they need.” 📈 And management isn’t

$Broadcom(AVGO)$ AI numbers are getting hard to ignore. AI semiconductor revenue hit $16.7B in Q3, up 221% year over year and 54% from the previous quarter. Management expects that figure to reach $21.7B in Q4.  But there’s another shift happening underneath those numbers. Semiconductor solutions now generated 70% of Broadcom’s total Q3 revenue, compared with 57% a year earlier. Infrastructure software was still growing, but at a much slower 29%.  That means Broadcom’s growth mix is changing quickly. The interesting question isn’t simply whether custom AI chips can compete with GPUs. It’s whether Broadcom can build a durable business around customers designing increasingly specialised AI infrastructure. And the cash flow is worth watching.

$Meta Platforms, Inc.(META)$ gained 3.24% Tuesday, but the interesting part isn’t the daily move. What caught my attention is how much expectation is now being built into Meta’s AI strategy. The company is spending aggressively on data centres, computing power and AI talent while simultaneously trying to keep its core advertising machine growing. That’s a very different investment equation from a company simply adding an AI feature to an existing product. Meta has something valuable that many AI companies don’t: billions of users and a huge advertising business that can potentially benefit from better recommendation systems, targeting and engagement. The challenge is turning that advantage into returns that justify the enormous investment. If AI i

The Anthropic filing puts a massive number on the table: agreements with $SpaceX(SPCX)$   worth up to $84.5B through 2029 for computing capacity. But I’m less interested in the headline number and more interested in what it tells us about SpaceX’s business. SpaceX is no longer just a rocket company. Starlink gives it a recurring connectivity business. Its launch operation provides access to orbit. And now its infrastructure is becoming part of the huge amount of computing capacity being secured by AI companies. That’s an interesting combination. The key question for me is whether these businesses can eventually reinforce each other rather than simply being separate projects under the same company. More Starlink capacity can support

One of the biggest questions for investors right now is what happens if interest rates stay higher for longer than the market expects. When rates are high, the investment landscape changes. Cash and short-term fixed income suddenly offer meaningful yields, borrowing becomes more expensive, and highly valued growth stocks can face more pressure as investors reassess what future earnings are worth today. But higher rates don’t necessarily mean sitting on the sidelines. If I had $10,000 to invest today, I’d be thinking about balancing three things: income, quality and flexibility. 🇺🇸 U.S. stocks I would still want exposure to equities, but I’d be more selective. Companies with strong balance sheets, consistent cash flow and pricing power can be better positioned if financing costs remain elev

Why $Micron Technology(MU)$ today? • Earnings are due September 30. • The market is watching whether the memory upcycle can continue at its current pace. • HBM demand remains a key driver as AI infrastructure spending expands. • MU has already had a huge run, so the interesting question isn’t simply “will earnings be good?” — it’s whether the numbers and guidance are good enough to justify the valuation.  The risk: expectations are already high. A strong report doesn’t necessarily mean the stock rises if guidance or margins disappoint. My angle: MU is becoming a test of whether the memory boom has more room to run — or whether expectations have simply moved too far ahead of fundamentals.

$NVIDIA(NVDA)$ New Test Isn’t AI Demand — It’s Expectations Nvidia has become so closely associated with the AI trade that almost every move in the stock gets interpreted through the same lens: AI demand. But Monday’s action raises a different question. When the broader semiconductor group sells off sharply while Nvidia holds up, perhaps the market isn’t simply choosing Nvidia as the “winner.” It may be separating companies with strong execution from companies where expectations have become harder to justify. That distinction could become increasingly important. Nvidia has built an unusually strong position in accelerated computing, but the stock also carries extremely high expectations. At this size, simply delivering good results may not be enou

$Direxion Daily Semiconductors Bull 3x Shares(SOXL)$ Is Down 6% — But Is the ETF Making the Selloff Look Worse? A 6% drop in SOXL looks dramatic, but there’s another way to read the move. SOXL is a 3x leveraged semiconductor ETF, which means investors aren’t just betting on chips — they’re betting on the daily direction of the sector with the move amplified. When semiconductor stocks pull back, SOXL can make an ordinary sector correction look much more extreme. That distinction matters right now. Semiconductor stocks are facing several pressures at once. Treasury yields have climbed sharply, with the U.S. 10-year recently moving above 5.2%, while oil prices are adding to inflation concerns. At the same time, investors are becoming more selective a