To The Moon

Earlier this year, one of Wall Street’s favorite bearish narratives was the “SaaSpocalypse.” The logic was simple: if AI agents can write code, build apps and automate workflows, why would companies keep paying large recurring fees to Salesforce, ServiceNow and other software vendors? That fear hit the sector hard. The S&P 500 Software & Services Index fell more than 26% from late January to its April low. Now the story is starting to reverse. The software index has climbed to a new 2026 high, and earnings expectations are moving higher as well. LSEG data shows expected 2026 earnings growth for the software sector has risen from about 13.8% at the end of March to 20.6%. The key shift is that AI is starting to look less like an immediate replacement for SaaS — and more like a new mo

A new pattern is emerging across the AI infrastructure boom: the companies selling compute are starting to help finance the companies buying it. According to Anthropic’s IPO filing, Broadcom has agreed to provide up to $42 billion in financing to support Anthropic’s AI infrastructure buildout. The financing could cover roughly one-third of Anthropic’s five-year $125.2 billion TPU lease commitment and may include convertible instruments that could eventually become equity. The headline number is huge, but the structure matters even more. Broadcom is already deeply involved in Google’s TPU ecosystem and is expected to help Anthropic access roughly 3.5GW of next-generation TPU capacity starting in 2027. Now Broadcom is also helping Anthropic finance that infrastructure. So the relationship is

HDD stocks took a sharp hit in the latest session. $Seagate Technology(STX)$ and $Western Digital(WDC)$ both fell about 10.2%, even as the broader tech market held up much better. This was not a broad AI hardware selloff. The pressure was concentrated in HDD names, and the trigger came from Toshiba. Toshiba reportedly plans to invest around ¥60 billion to expand its Philippine operations and roughly double HDD capacity for AI data centers by fiscal 2027 compared with 2025 levels. The company is also targeting a much larger share of the global HDD market over time. That immediately raised a key concern for investors: if the industry’s No. 3 player starts adding meaningful capacity, how long can Seagate and W

Memory stocks have been one of the hottest parts of the AI trade, but Michael Burry is leaning the other way. Burry has continued to add to his bearish exposure on $Micron Technology(MU), even as DRAM pricing remains firm and AI-related demand stays strong. What makes the trade interesting is the timing: he is not shorting memory because the current fundamentals look weak. He appears to be betting that today’s strength eventually creates tomorrow’s oversupply. That is the core debate in memory right now. The bullish case is straightforward. AI servers need more HBM, more server DRAM and more enterprise SSD capacity. Hyperscalers are still expanding infrastructure, memory content per server keeps rising, and supply remains tight in several key categories. In that environment, strong pricing

Apple’s latest Mac update is about more than faster hardware. The new Mac mini and Mac Studio are being positioned as machines that can run AI agents, large language models and enterprise workflows locally. The high-end Mac Studio can support up to 512GB of unified memory, while multiple Macs can be linked together for distributed inference. Apple has even demonstrated four Mac Studios running a trillion-parameter model using a standard wall outlet. The more interesting part is how Apple is selling the economics. Cloud AI usually charges by usage. The more tokens a company consumes, the more it pays. Apple’s pitch is different: buy the hardware once, then keep running workloads locally without paying for every model call. That matters when AI usage becomes frequent. If an enterprise agent

AI may be moving from “answering questions” to actually doing work — and that could make CPUs important again. The latest AI trade is no longer just about GPUs. Meta’s new AI agent, Muse, has pushed investors to rethink what the next phase of AI infrastructure may actually require. Why?Because an AI agent does much more than generate an answer. It may need to:open a browser,search the web,fill out forms,call APIs,run tools,manage files,and keep working in the background for minutes or even hours. That changes the compute equation. For a chatbot, the workflow is relatively simple: Prompt → GPU inference → Response For an AI agent, it looks more like: Think → Browse → Execute → Check → Think Again → Continue GPUs still handle the heavy model inference. But many of the surrounding workloads —

3.2T and 6.4T solutions are showing up at ECOC 2026. The next AI bottleneck may be shifting from raw compute to connectivity. As AI clusters scale from thousands to hundreds of thousands of GPUs, one problem is becoming increasingly important: How do you move massive amounts of data between those GPUs fast enough — and without burning too much power? That is exactly why optical interconnects are becoming a bigger part of the AI infrastructure story. At ECOC 2026, Coherent showcased a 3.2T OSFP optical module as well as a 6.4T NPO optical engine designed for next-generation AI scale-up and scale-out networks. The direction is becoming clear: 800G → 1.6T → 3.2T And the upgrade is not just about higher headline speeds. The bigger shift is that optics are moving closer to the chip. As GPU dens

The Bank of Japan has raised its policy rate by 25 basis points to 1.25%, the highest level in 31 years. The move passed by a 7-2 vote and was broadly expected by markets. The bigger question now is not the 1.25% level itself, but how far the BOJ is prepared to go from here. This matters far beyond Japan. For years, the yen has been one of the world’s cheapest funding currencies. Investors could borrow at very low Japanese rates and move that capital into higher-yielding assets elsewhere — U.S. stocks, bonds, emerging-market currencies and other risk assets. That is the basic logic behind the yen carry trade. As Japanese rates rise, that trade becomes less attractive. If the yen also strengthens, investors face both higher funding costs and FX losses. That is why every BOJ tightening cycle

One of today’s more interesting semiconductor stories is not about a new GPU or a new AI model. Reuters reported that SK hynix is in exploratory talks with Intel about producing memory chips in the U.S. for the first time. One option under discussion is for SK hynix to use part of Intel’s Ohio fab capacity. Another possibility is a joint structure involving SK hynix, Intel and potentially major cloud customers. The talks are still at an early stage, and there is no final decision yet on product scope, investment size or structure. What makes this interesting is that this is not simply another “chipmaker builds in America” story. SK hynix already has a U.S. footprint, including its advanced AI-memory packaging project in Indiana. If front-end memory production also moves closer to U.S. cust

The Fed decision tonight is important, but the market may already have moved beyond the first question. A 25bp hike is now largely priced in, which means the bigger issue is no longer simply “Will the Fed hike?” but “Does this mark the start of another tightening cycle, or is it just a one-off adjustment?” If the Fed raises rates by 25bp as expected, the target range would move higher again, but the market reaction will likely depend much more on the new dot plot and the tone of the press conference than on the headline rate move itself. The reason expectations shifted so quickly is that the latest inflation data have remained uncomfortable while the labor market has not weakened enough to give the Fed much room to ignore it. CPI and PPI both showed renewed price pressure, while payroll gr