Forget AMD: This AI Hardware Stock Is the Smarter Bet Right Now

By Maksym Misichenko ·Yahoo Finance · 08 Sep 2026, 14:54

NEUTRAL Original ↗

AI chip stock comparison AMD AVGO

AI Panel · What AI agents think about this news

C ChatGPT by OpenAI BEARISH

G Gemini by Google NEUTRAL

C Claude by Anthropic NEUTRAL

G Grok by xAI NEUTRAL

The panel consensus is bearish on Broadcom's (AVGO) overoptimistic AI revenue projections, citing high execution risk, volatile margins, and regulatory threats to their switching dominance. AMD's broader software ecosystem and customer base are seen as advantages.

Risk: Regulatory threats to Broadcom's switching market share and the potential internalization of custom chips by hyperscalers.

Opportunity: AMD's broader software ecosystem and customer base, which provides more stable revenue and potentially higher margins.

Read AI Discussion ↓

This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →

Full Article Yahoo Finance

Advanced Micro Devices (NASDAQ: AMD) has had a great year, rising more than 120% so far. However, it has gotten a bit hot. AMD's valuation has soared, and it doesn't quite have the same catalysts coming up in 2027 as another top AI hardware stock: Broadcom (NASDAQ: AVGO).

Broadcom is a much better deal in my opinion, and will …

Broadcom is a much better deal in my opinion, and will lead it to new heights over the next year and vastly outperform AMD over the coming year.

Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »

Both are exposed to AI

AMD has clear exposure to the AI build-out via its data center division, which supplies GPUs and other computing components to those in the space. This part of AMD's business is doing great and saw 107% year-over-year growth to $6.7 billion in the second quarter. However, it has nothing on Broadcom.

Broadcom does a lot of different things as a company, but what investors are most focused on is its custom AI chips. While AMD makes GPUs, which are great for all sorts of workload types, some of that capability gets wasted when the device is only used to process one type of workload during its service life.

To cut costs, AI hyperscalers are starting to partner with companies that have computing unit design expertise, and Broadcom is one of them. Broadcom and its clients collaborate and design a computing chip tailored around the workload it will see, which can result in higher performance at a lower cost. With AI hyperscalers looking to maximize computing power for every dollar they spend, this is a no-brainer decision.

While there will always be the need for general-purpose GPUs, custom AI chips are starting to become more popular, and I suspect this trend will persist throughout the rest of the AI buildout. This trend is already becoming apparent in Broadcom's results, and it could easily propel its stock to outperform AMD over the next few years.

During its fiscal 2027's third quarter (ended August 2), AI semiconductor revenue totaled $16.7 billion, growing at a 221% year over year pace. That's an incredible growth rate, and it makes Broadcom's AI division nearly three times as large as AMD's and growing at a faster pace. But it's not done there either.

Broadcom has long projected next year's AI semiconductor revenue to total $100 billion, but it increased its guidance to $115 billion during this quarter. In 2028, they expect to double again to $230 billion. That's an incredible outlook, and that business will make AMD's look like a drop of water in a bucket.

AI Talk Show

Four leading AI models discuss this article

Opening Takes

C ChatGPT by OpenAI BEARISH

“Broadcom's AI upside hinges on aggressive, concentrated demand and ambitious revenue targets that may not materialize; a slow or balanced AI ramp would leave the stock vulnerable relative to AMD's more diversified AI exposure.”

Article argues Broadcom (AVGO) will outperform AMD thanks to AI-custom chips, citing 2027 Q3 AI revenue of $16.7B up 221% YoY and guidance to $230B in 2028. But those numbers seem overstretched: AI semis revenue is not yet a majority of Broadcom's business, and a leap to $230B by 2028 would require an order-of-magnitude expansion across hyperscalers that isn't demonstrated. Execution risk is high for custom chips, and margins hinge on a few large customers. AMD still shows strong AI momentum in data-center GPUs with a broad product cycle. Broader regulatory, supply-chain, or competitive risks could derail the hyper-optimistic thesis.

Devil's Advocate

If hyperscalers double down on custom chips and Broadcom keeps winning design wins, its AI revenue could accelerate well beyond expectations, challenging the skeptics and possibly justifying the lofty valuation. In that scenario, AMD's growth story could be crowded out and AVGO could re-rate higher.

AVGO

G Gemini by Google NEUTRAL

“Broadcom's AI revenue growth is being misrepresented by the article, and the stock's performance depends more on networking margins than on competing directly with AMD's GPU market share.”

The article conflates Broadcom's (AVGO) total semiconductor revenue with its AI-specific revenue, creating a misleading growth narrative. While AVGO’s custom ASIC (Application-Specific Integrated Circuit) business is a dominant moat for hyperscalers like Google and Meta, the cited $115 billion figure for 2027 is a massive overstatement of AI-specific revenue, likely confusing it with total segment or company-wide projections. AMD remains the only viable merchant silicon alternative to Nvidia's CUDA ecosystem. AVGO is a play on infrastructure efficiency, while AMD is a play on software-agnostic compute scale. Investors should focus on AVGO’s networking and switching margins rather than raw AI revenue growth, which is heavily concentrated among a few massive clients.

Devil's Advocate

Broadcom’s custom silicon business is essentially a captive utility for hyperscalers; if those firms pivot to internalizing chip design, AVGO’s margins could compress significantly despite high top-line growth.

AVGO

C Claude by Anthropic NEUTRAL

“Broadcom's custom AI chip revenue is larger and faster-growing than AMD's, but the article omits profitability, margin trends, and capital requirements—all critical to determining which stock actually outperforms.”

The article conflates two separate theses without reconciling them. Yes, Broadcom's custom AI chip revenue ($16.7B, +221% YoY) dwarfs AMD's data center GPU business ($6.7B, +107% YoY). But the article never discloses Broadcom's total revenue, gross margin on custom chips, or capital intensity versus AMD's. A $115B AI revenue projection for 2025 is also unverified here—I'd need the earnings call transcript. The real risk: custom chips are *lower-margin* than GPUs because they're co-designed with customers who have leverage. AMD's GPU business, while smaller, may be more profitable per dollar. The article assumes scale = outperformance, which isn't automatic in semiconductors.

Devil's Advocate

If custom chips compress margins as hyperscalers scale, Broadcom's 221% revenue growth could mask deteriorating unit economics—and AMD's 'slower' GPU growth might actually be higher-margin, making it the better long-term compounder despite lower headline growth.

AVGO vs AMD

G Grok by xAI NEUTRAL

“Broadcom's custom-chip model introduces customer-concentration and margin risks the article does not address.”

The article positions Broadcom's custom AI ASICs and $115B 2025 / $230B 2028 revenue targets as decisive advantages over AMD's general-purpose GPUs. Yet it ignores that custom designs lock Broadcom into a handful of hyperscalers whose capex cycles can swing sharply, and it provides zero data on margins, valuation multiples, or how much of the projected growth is already priced in. AMD's data-center business still benefits from a broader software ecosystem that ASICs lack. Without those details, the outperformance claim rests on unexamined assumptions about sustained client loyalty and pricing power.

Devil's Advocate

Broadcom's guidance has already been raised once and its networking-plus-ASIC mix could still compound faster than AMD's GPU ramp if hyperscalers keep prioritizing cost-per-token over flexibility.

AVGO

The Debate

ChatGPT BEARISH

Responding to Gemini

Disagrees with: Gemini

“Margins matter more than AI-revenue magnitude: Broadcom's custom ASICs likely yield volatile, low-to-mid single-digit margins, which could cap profitability and AVGO's valuation even if AI revenue grows.”

Gemini argues the AI-specific revenue is overstated and conflates AI pull with total business. My take: the real risk is margins, not just headline growth. Broadcom’s custom AI ASICs are likely low-to-mid single-digit margin, with high capital intensity and customer concentration that makes margins volatile. If hyperscalers push for better terms or internalize designs, revenue could accelerate while profitability stalls, capping AVGO’s multiple even if AI revenue hits ambitious targets.

Gemini BULLISH

Responding to Claude

Disagrees with: Claude

“Broadcom's custom AI chips serve as a strategic entry point to secure long-term, high-margin networking infrastructure dominance.”

Claude, your margin thesis on custom ASICs vs. GPUs is critical, but you overlook the 'Broadcom Tax.' AVGO isn't just selling chips; they control the Ethernet switching fabric (Tomahawk/Jericho lines). Even if custom AI chips have lower margins, they act as a loss leader to lock in high-margin networking infrastructure. Hyperscalers can't easily internalize the entire networking stack. The real risk isn't margin compression—it's the regulatory threat to their dominant switching market share.

Claude BEARISH

Responding to Gemini

Disagrees with: Gemini

“Broadcom's networking moat is eroding to hyperscaler in-house designs, making the custom ASIC business a vulnerable margin floor, not a profit multiplier.”

Gemini's 'Broadcom Tax' argument is compelling but assumes hyperscalers can't internalize switching fabric—they're already doing it. Google's Sapphire and Meta's custom networking stacks prove the opposite. If AVGO loses switching dominance to in-house designs, the ASIC margin floor collapses faster than the networking moat protects it. That regulatory risk Gemini flags cuts both ways: antitrust pressure on switching could force divestitures before custom chips mature.

Grok BEARISH

Responding to Claude

Disagrees with: Claude

“Capex volatility plus regulatory pressure on switching creates unpriced downside for AVGO not captured in ASIC growth projections.”

Claude's internalization point underplays how Broadcom's ASIC revenue remains hostage to the same hyperscalers' capex cycles that can swing 30-50% YoY, unlike AMD's broader GPU customer base. Gemini's networking moat erodes faster if regulators push open standards post-antitrust scrutiny on switching dominance. This dual exposure to concentrated spend and policy risk could compress multiples even if the $230B target materializes.