U.S. AI stocks are under pressure in Friday’s premarket as Broadcom drops on a margin outlook, and Oracle’s capex surge reignites AI-bubble fears.
NEW YORK — Friday, December 12, 2025 (Premarket, ~6:00 a.m. ET)
U.S. equity futures are pointing to a mixed open Friday morning, with the AI trade again lagging after two back-to-back “reality check” earnings reactions: Oracle’s spending surge and Broadcom’s margin warning. As of about 5:47 a.m. ET, Nasdaq futures were down ~0.47%, S&P 500 futures down ~0.11%, while Dow futures were up ~0.22%—a setup that underscores the market’s growing preference for “profits now” over “promise later” in parts of AI. [1]
Below is what’s moving AI-related U.S. stocks before the bell, what Wall Street is debating, and the key catalysts to watch through today’s session.
Premarket snapshot: what’s moving AI stocks right now
The theme: Investors are not questioning AI demand. They’re questioning margins, cash flow, and financing—especially where the AI buildout requires massive capex.
Key movers and pressure points (early premarket):
- Broadcom (AVGO): Shares are down around 5% after the company forecast strong revenue but flagged a near-term margin dip tied to its AI business mix. [2]
- Oracle (ORCL): After Thursday’s sharp selloff tied to guidance and spending, Oracle remains a focal point for the market’s “AI capex payback” debate. [3]
- Nvidia (NVDA), AMD (AMD), Micron (MU) and the chip complex: These names are reacting to the same story: if hyperscalers and cloud challengers keep spending, chip demand stays hot—but investors are growing sensitive to who captures the profits. [4]
Why the AI trade is wobbling: it’s not demand—it’s “who pays” and “who profits”
The AI rally of 2025 has increasingly split into two narratives:
- AI demand is real (training, inference, enterprise rollouts, and data-center buildouts).
- AI economics are messy—especially for firms funding infrastructure with heavy borrowing, or for suppliers whose AI mix pressures margins.
That’s why this week’s market action matters: Oracle and Broadcom sit on opposite sides of the AI buildout—buyerversus supplier—and both just revealed reasons investors are turning more selective.
Broadcom earnings: bullish revenue forecast, bearish margin message
Broadcom is a bellwether because it’s positioned at the heart of AI infrastructure: networking, custom silicon (ASICs), and the broader data-center stack.
What Broadcom forecast (the bullish part)
Broadcom projected first-quarter revenue of about $19.1 billion, ahead of Wall Street estimates around $18.27 billion. [5]
It also said AI semiconductor revenue is expected to double to about $8.2 billion in the upcoming quarter—an eye-catching sign that AI demand remains strong into early 2026. [6]
Broadcom’s CEO also highlighted a $73 billion order backlog scheduled for delivery over the next 18 months, concentrated among roughly five customers—a detail investors watch closely because it speaks to both visibility and customer concentration risk. [7]
Why the stock fell anyway (the margin/mix problem)
Broadcom warned consolidated gross margin could fall by roughly 100 basis points sequentially, and pointed to a core driver: a rising revenue share from AI chip products and systems that can carry lower margins and higher investment costs. [8]
In plain English:
AI revenue is growing fast—but it may be diluting profitability in the near term.
That’s a problem for a market that’s recently rewarded AI suppliers for “picks-and-shovels” economics (high margins, strong pricing power). Today’s reaction suggests investors are increasingly asking for proof of durable AI margins, not just AI growth.
The bigger implication for the AI hardware basket
Broadcom’s results also matter because it’s one of the leaders in custom AI chips (ASICs) built for hyperscalers—often framed as an alternative (or complement) to Nvidia GPUs. The market’s question is shifting from “Will custom silicon happen?” to “Will it pressure margins across the stack?”
And it’s happening as U.S. cloud providers are expected to spend more than $400 billion on AI this year, reinforcing that the spend cycle is huge—but not automatically profitable for every participant. [9]
Oracle: the AI capex gut-check that reignited “bubble” fears
Oracle is central to the debate because it has been pitching itself as a major AI cloud infrastructure player—especially after striking a headline-grabbing partnership with OpenAI.
The numbers that spooked the market
In its earnings outlook, Oracle projected:
- Fiscal Q3 adjusted EPS of $1.64–$1.68, below estimates around $1.72. [10]
- Revenue growth guidance that also came in below analyst expectations. [11]
- Capital expenditures for fiscal 2026 expected to be $15 billion higher than the roughly $35 billion figure discussed earlier—an enormous step-up that drew immediate scrutiny. [12]
Reuters Breakingviews framed the moment as a sign that investor enthusiasm for AI may be cooling—because the spending is arriving faster than the cash generation. [13]
“Chip neutrality” and financing models: Oracle signals flexibility (and pressure)
Oracle’s leadership also emphasized “chip neutrality”—essentially positioning Oracle as willing to deploy whatever chips customers want, while still buying Nvidia GPUs. [14]
More importantly for investors, Oracle discussed financing approaches that could reduce Oracle’s upfront capex burden—such as:
- Customers bringing their own chips, and
- Vendors exploring “rent capacity” models instead of outright selling hardware. [15]
Those comments matter because they imply the market is forcing a shift from “build everything ourselves” toward more creative funding structures.
Credit and cash-flow worries became part of the equity story
As Oracle’s spending plans ballooned, investors also focused on its debt load and credit risk signals. Reuters reported heightened investor concern about Oracle’s borrowing and rising costs to insure its debt against default after the results miss. [16]
This is unusual for a mega-cap tech name in an AI boom—and it’s a big reason the selloff spilled into the broader AI complex.
Read-through: what Oracle and Broadcom mean for Nvidia, AMD, and the chip ecosystem
The last 48 hours have effectively tested a market assumption:
“If AI capex rises, AI chip stocks always win.”
This week suggests the market is now adding conditions:
- Capex can help demand, but it can also scare equity investors if it looks debt-funded or inefficient.
- Suppliers can grow AI revenue, but the market wants to see margin durability, not dilution.
Nvidia: demand tailwinds—but policy and competition headlines add volatility
Beyond the capex debate, Nvidia is also contending with export-policy headlines.
Earlier this week, Reuters reported the U.S. would allow exports of Nvidia’s H200 chips to China under a framework involving a 25% fee, sparking renewed debate about national security vs. commercial competitiveness. [17]
A Reuters report also said Chinese firms including ByteDance and Alibaba were interested in ordering H200 chips after the U.S. green light—though approval from Chinese authorities and supply clarity remained key uncertainties. [18]
At the same time, Reuters also reported Beijing could limit access to the H200 chips despite U.S. approval, adding another layer of uncertainty about how much revenue actually materializes. [19]
And late Thursday, Reuters reported U.S. Senator Elizabeth Warren called for Nvidia CEO Jensen Huang and Commerce Secretary Howard Lutnick to testify over the China chip sales decision—signaling that the story could stay in the headlines and inject volatility into the stock. [20]
AMD: positioned for “multi-vendor” AI, but still traded as part of the AI basket
Oracle’s “chip neutrality” message can be read as supportive of AMD’s long-term positioning, given Oracle and AMD previously announced plans for a major AI supercluster deployment built around 50,000 AMD Instinct MI450 GPUsbeginning in calendar Q3 2026. [21]
That doesn’t mean AMD is immune to today’s sentiment; it still trades with the AI beta. But it does show that the AI market is not “one-chip-wins-all,” and investors may increasingly price a multi-vendor AI hardware future.
What to watch in today’s session: 5 catalysts driving AI stocks
1) Can Broadcom stabilize after the “margin” headline?
Watch whether investors treat the margin dip as a short-term mix issue—or as a signal that AI infrastructure is becoming more competitive and less profitable.
2) Will Oracle’s selloff cool, or spread further?
The key question is whether investors view Oracle’s capex surge as:
- a necessary land-grab to meet contract demand, or
- a warning sign that the sector is chasing growth at the expense of balance-sheet strength.
3) Does the market keep rotating away from AI into value?
Thursday’s market action showed a rotation pattern: the Dow and S&P 500 closed at records even as AI-heavy tech segments lagged. [22]
If that continues, AI stocks may struggle even if the broader market stays green.
4) Are export headlines a net positive or negative for Nvidia?
China-related chip headlines can shift quickly from “new demand” to “new restrictions” to “new political scrutiny.” This is a key driver of short-term volatility. [23]
5) The bigger question: AI spending is huge—who captures the profits?
A core tension is emerging: cloud buyers and chip suppliers are both spending heavily. The market is increasingly rewarding companies that can show:
- rising AI revenue and
- stable or improving margins and
- credible cash-flow paths.
Bottom line for AI stocks today (Dec. 12, 2025)
Premarket action suggests the AI theme isn’t “over”—but it is being repriced. The market is shifting from an AI story driven by enthusiasm to one driven by fundamentals:
- Broadcom is proving AI demand can surge while margins get squeezed.
- Oracle is proving AI capex can accelerate while investors worry about funding, cash burn, and credit.
- Nvidia and AMD remain long-term beneficiaries of AI compute demand, but sentiment is being shaped by both capex anxiety and policy headlines.
Reminder: Premarket moves can reverse quickly after the opening bell, especially around mega-cap tech and semiconductors.
References
1. seekingalpha.com, 2. www.reuters.com, 3. www.reuters.com, 4. www.reuters.com, 5. www.reuters.com, 6. www.reuters.com, 7. www.reuters.com, 8. www.reuters.com, 9. www.reuters.com, 10. www.reuters.com, 11. www.reuters.com, 12. www.reuters.com, 13. www.reuters.com, 14. www.reuters.com, 15. www.reuters.com, 16. www.reuters.com, 17. www.reuters.com, 18. www.reuters.com, 19. www.reuters.com, 20. www.reuters.com, 21. www.oracle.com, 22. www.reuters.com, 23. www.reuters.com
