JPMorgan Shuts Down Broadcom Delay Fears — TPU v9 Is Right on Track
Why Billionaire Dan Loeb and Wall Street Are Bullish on Broadcom (AVGO)
Wall Street Analyst Pushes Back on Delay Worries
Imagine you’re building a really cool treehouse, and suddenly people start saying, "Oh no, they’re way behind schedule!" But then your lead builder steps in and says, "Not true — everything is right on track." That’s basically what happened here.
On June 17, Harlan Sur, an analyst at the big bank JPMorgan, reaffirmed a strong vote of confidence in Broadcom Inc. (NASDAQ: AVGO). He maintained an Overweight rating — which is Wall Street’s way of saying, "We think this stock is a good bet" — and set a $580 price target, meaning he believes the stock price could climb significantly.
But here’s the exciting part: Sur used this moment to shut down growing fear that Broadcom’s work on an AI chip with Google (called the TPU v9 program) had run into serious problems.
What Were People Worried About?
Rumors were floating around that Broadcom’s TPU v9 — a next-generation artificial intelligence (AI) accelerator being built for Google — was facing significant delays. People were getting nervous. An AI accelerator is basically a special type of computer brain designed to handle the enormous amount of number-crunching that AI requires.
What Did the Analyst Say?
Harlan Sur said those fears were not true. He confirmed that:
- The TPU v9 program remains one of Broadcom’s highest-priority projects
- The project is firmly on schedule for a 2028 production ramp (that’s when they start making it in large quantities)
- There are no delays or cancellations confirmed
Important Point: When a major Wall Street analyst "pushes back" against negative rumors like this, it’s a strong signal that the company’s plans are still moving forward. This kind of news can reassure investors and sometimes gives the stock price a boost.
So, What Exactly Is the TPU v9 Chip?
Let’s break this down like a recipe. The TPU v9 chip is an incredibly advanced piece of technology. Here’s what’s inside it:
- 4 compute dies — Think of these as the four "brains" doing the actual thinking and calculating
- 16 High Bandwidth Memory (HBM) stacks — These are like super-fast notepads that hold information the chips need right now, so they don’t waste time looking elsewhere
- 400Gbps SERDES — These are ultra-high-speed data highways that let different parts of the chip (and different chips) talk to each other incredibly fast
All of this is built using 2nm (nanometer) technology, which means the tiny electrical pathways inside the chip are unbelievably small — allowing for more power and efficiency. The whole package is designed to handle the massive data demands of next-generation AI workloads — think smarter chatbots, better image recognition, and more powerful AI in general.
How Is the Project Progressing? A Timeline
Broadcom’s work with Google on AI chips has been moving forward in clear steps:
- First half of 2025: Broadcom completed the initial IP design phase — this is like finishing the blueprints
- Second half of 2025: The team moved into full system-on-chip (SoC) design — now they’re actually building the detailed plans for the full chip
- 2028 (projected): The chip is expected to enter production ramp — meaning manufacturing begins at scale
The Bigger Picture: Broadcom isn’t new to working with Google. They’ve been collaborating on AI chips across multiple generations, and each project builds on the lessons learned from the last one.
What About the Current Generation Chip? The TPU v8i
While all the attention might be on the future v9 chip, the current generation is already making big moves:
- The TPU v8i (3nm chip) received full qualification from Google by mid-2025
- It’s set to begin production ramping this quarter — meaning Google has tested it, approved it, and is now ready to start making lots of them
This is a huge deal because qualification means Google has given the chip a big thumbs-up after rigorous testing.
How Does Broadcom Compare to Google’s Internal Team?
This is where it gets really interesting. Google also has its own internal chip team working on a competing program called the Zebrafish TPU v8t (3nm), in partnership with a company called MediaTek.
Here’s the key difference:
| Broadcom (TPU v8i) | Google/MediaTek (Zebrafish TPU v8t) | |
|---|---|---|
| Status | Fully qualified, production ramping | Still in design optimization phase |
| Timeline | Ahead by a lot | Still working on improvements |
That means Broadcom is approximately 18 months ahead of Google’s own internal team. That’s a massive lead in the fast-moving world of technology — like being a full school year ahead of your classmates in a subject.
What Does Broadcom Actually Do?
For a quick overview, Broadcom Inc. is a semiconductor and infrastructure software company. In plain English:
- Semiconductor side: They design and build tiny computer chips used in data centers (the huge buildings full of computers that power the internet), broadband connections, wireless communications, and storage systems
- Software side: They provide mainframe software and cybersecurity solutions — basically, they help keep big computer systems safe and running smoothly
So Broadcom isn’t just an AI play. It’s a diversified company with its hands in many important technology areas.
Important Point: Diversification means Broadcom isn’t betting everything on one type of product. Even if one area slows down, other areas can help keep the company growing.
Should You Invest in Broadcom?
The article notes that while Broadcom is a solid company with strong potential, some AI stocks may offer greater upside (bigger rewards) with less downside risk. The authors hint at an undervalued AI stock that could benefit from Trump-era tariffs (taxes on imported goods that help U.S. companies) and the onshoring trend (companies bringing manufacturing back to the United States).
They point readers to a free report on what they consider the best short-term AI stock for more details.
Summary
- Broadcom (AVGO) is considered one of the best growth stocks by billionaire investor Dan Loeb
- JPMorgan analyst Harlan Sur confirmed an Overweight rating with a $580 price target on June 17
- Sur pushed back against rumors that Broadcom’s AI chip project with Google (TPU v9) was delayed — confirming it’s on schedule for 2028
- The TPU v9 chip features advanced specs: 4 compute dies, 16 HBM stacks, and 400Gbps SERDES on a 2nm process
- The current-generation TPU v8i chip is already qualified and entering production, giving Broadcom an 18-month head start over Google’s rival internal program
- Broadcom is a diversified semiconductor and software company with products in data centers, broadband, wireless, storage, and cybersecurity
Frequently Asked Questions
What is Broadcom’s relationship with Google?
Broadcom has been Google’s key partner in designing custom AI accelerator chips called TPUs (Tensor Processing Units) across multiple generations. They work together to build the specialized hardware that powers Google’s AI services.
What does "TPU v9 2nm" mean?
TPU v9 is the ninth generation of Google’s custom AI chip. The 2nm refers to the manufacturing process — the smaller the number, the more tiny transistors can fit on the chip, making it more powerful and energy-efficient.
What does an "Overweight" rating from JPMorgan mean?
It’s Wall Street’s way of saying the analyst thinks the stock will outperform the overall market. It’s a green light suggesting investors should consider buying or holding the stock.
Why is Broadcom’s 18-month lead over Google’s internal team significant?
In the fast-paced tech industry, 18 months is an eternity. Being nearly a year and a half ahead of a competitor means Broadcom can start selling chips, earning revenue, and refining their technology long before anyone else catches up. This lead can translate into a stronger market position and more profits.
What risks does Broadcom face despite the positive analyst report?
No investment is risk-free. Potential risks include changes in AI spending by big tech companies, competition from other chipmakers, and broader economic factors that could slow down technology investments. The article itself acknowledges that other AI stocks might offer even higher potential rewards.
