The AI Trade Is Moving Past Chips Into Power, Services and Cash Flow

The pick-and-shovel version of the AI trade is broader than most portfolios reflect, stretching from enterprise workflow automation to fuel cells sitting on the hyperscaler power bottleneck.

You also get a cash-return compounder where buybacks and margin expansion do most of the heavy lifting. Each name has a catalyst inside the next 60 days.

AI wealth (Sponsored)

The Next Trillionaires Are Making the Same Bet

Some of the world's most influential tech leaders are investing heavily in what many consider the next phase of artificial intelligence.

According to one analyst, this shift could create new winners and losers across the stock market.

He's released a free Hotlist and Hitlist highlighting the stocks he believes may be best positioned—and those that may face the greatest challenges.

Get the Free Hotlist + Hitlist

Graphics

Qualcomm Brings AI Rendering Into the GPU Pipeline

Qualcomm (NASDAQ: QCOM) is adding Adreno Neural Fusion to its graphics architecture, combining AI super-resolution, frame generation, and neural processing directly inside the GPU pipeline. New Adreno Matrix Cores handle AI workloads, while local graphics memory keeps more processing close to the chip.

The approach lets the GPU use AI while frames are being created, rather than treating neural features as a separate step afterward. That can improve image quality and responsiveness without sending the work elsewhere.

Frame Generation Gets Dedicated Hardware

Adreno Neural Fusion can create additional frames between traditionally rendered ones while AI upscaling sharpens lower-resolution images. Both techniques aim to raise visual quality and frame rates without requiring the GPU to render every pixel from scratch.

Qualcomm is also using 18MB of local graphics memory to reduce data movement inside the system. Keeping frequently used information closer to the graphics hardware can lower latency and improve efficiency during heavier workloads.

Game Engines Plug Straight Into the Stack

Support for Unity and Unreal Engine gives developers access to the new architecture through two of the most widely used game engines. Studios can add AI-assisted rendering without building an entirely separate graphics pipeline.

Qualcomm is making neural processing a native part of mobile graphics rather than an optional layer on top. Adreno is evolving from a traditional GPU into a hybrid graphics engine where rasterization and AI work together in real time.

Google Cloud

Google Builds a Faster AI Model for Real-World Weather

Alphabet (NASDAQ: GOOGL) is rolling out WeatherNext 3, a new AI model built to generate weather forecasts faster and more frequently. The system is designed to predict changing conditions across temperature, wind, precipitation, and other atmospheric variables while requiring far less computing time than traditional forecasting methods.

Google plans to feed those predictions directly into products people already use. Search, Maps, and Gemini will all gain access to the new forecasting layer.

Millions of Forecasts, Much Less Waiting

WeatherNext 3 can generate large groups of possible weather outcomes instead of relying on a single prediction. Comparing those scenarios helps the system estimate uncertainty and better capture events where conditions can shift quickly.

The model can also produce forecasts more often, giving Google fresher information as new weather data arrives. Developers and researchers will be able to access the technology through Google Cloud for their own applications.

AI Steps Into Physical-World Prediction

Weather forecasting gives Google AI a very different job from generating text, images, or code. The model has to process enormous amounts of atmospheric data and predict how real conditions will develop over time.

By connecting WeatherNext 3 with Maps, Search, Gemini, and Cloud, Google can turn one forecasting system into infrastructure used across consumer and enterprise products. AI is becoming part of the machinery behind everyday information, not simply an assistant sitting on top of it.

A-Rated AI Opportunity (Sponsored)

My stock-grading system just flagged this AI company

Louis Navellier says his $9 million Stock Grader system is flashing its highest rating on one AI stock for 2026.

This company recently reported record annual revenue of $113.5 billion, earned an A-rating in his system, and is tied to a major AI expansion linked to Elon Musk’s “Project Apex.”

Navellier says this is the same system that helped him spot Nvidia long before its historic run.

Now, he is revealing the name, ticker, and full analysis of his #1 AI stock pick for 2026, completely free.

Click here to get Louis Navellier’s #1 AI stock for 2026 before this free briefing comes down.

AI Workstations

Microsoft Launches Project Zenith for On-Device AI Development 

Microsoft (NASDAQ: MSFT) is launching Project Zenith, a Windows 11 experience designed for PCs powerful enough to run large AI models directly on the device. The first systems will pair AMD Ryzen AI Halo hardware with at least 64GB of unified memory and more than 250GB/s of memory bandwidth.

That hardware gives developers enough room to work with models containing more than 30 billion parameters without sending every request to remote cloud servers. Coding assistants, experiments, and agent workflows can stay on the machine.

Memory Becomes the AI Advantage

Unified memory lets the CPU, GPU, and AI hardware work from the same large pool of data instead of constantly moving information between separate memory systems. For local models, that can make a major difference in how smoothly large workloads run.

Microsoft is also shipping the experience ready for development, with Windows Terminal and Visual Studio Code configured from the start. Project Zenith is meant to feel like an AI development machine rather than a regular PC that needs hours of setup.

Agents Get a Safer Place to Work

Windows security technology is part of the design as well. Microsoft Execution Containers can isolate AI agents while they run code or interact with local tools, helping limit what automated systems can access.

Project Zenith gives Windows a new role in the AI stack. Microsoft is building a PC platform where models, agents, development tools, and security can all operate locally under one system.

Tax Strategy (Sponsored)

3 Overlooked Ways to Potentially Reduce Capital Gains Tax

Capital gains taxes can take a bigger bite out of your profits than expected.

Fortunately, some deductions may help reduce the impact — including:

• Investment-related expenses

• Cost basis adjustments

• Certain real estate selling costs

Because rules and eligibility vary, many investors turn to fiduciary financial advisors for guidance.

Find an Advisor Match.

Everything Else

• 📊 A rate cut looks increasingly likely heading into fall. See the 10 stocks set up to respond first as the next phase of the cycle begins.

• 📶 Nokia is set to rejoin the Euro STOXX 50, returning to Europe’s blue-chip index as Volkswagen drops out.

• 🧠 Samsung Electronics and SK Hynix are part of new U.S. chip talks as South Korea negotiates semiconductor investment terms ahead of planned U.S. tariffs.

• 🏭 Wacker Chemie is weighing whether to close its Tennessee plant after new U.S. trade measures cost the polysilicon facility its final two customers.

• 🖥️ Hut 8 is developing the Texas data center behind Anthropic’s $35 billion cloud deal with Lambda as AI infrastructure spending continues to accelerate.

• 💻 Salesforce raised its annual forecasts and expanded its AI partnership with Anthropic as demand for enterprise AI tools continues to grow.