STOCK MARKETS

5 big analyst AI moves: Nvidia hit with downgrades after Huang’s exit

By Investing.com8 min readInvesting.com
5 big analyst AI moves: Nvidia hit with downgrades after Huang’s exit

Here are the biggest analyst moves in the area of artificial intelligence (AI) for this week. 

Analysts downgrade Nvidia after Huang’s exit

Nvidia CEO Jensen Huang has resigned, the company said in a regulatory filing, not long after media reports revealed that its board was preparing to remove him. CFO Colette Kress has been appointed interim CEO, and Harvey C. Jones has been named the board’s first-ever chair.

Wall Street analysts downgraded Nvidia and cut price targets, adding to the pressure on a stock already weighed down by the leadership turmoil.

On Wednesday, Reuters reported that Huang had an intimate relationship with a second employee, one who reports to him. The reports followed earlier news of Huang’s relationship with Alice Huang, an executive at Nvidia customer Megaspeed. U.S. intelligence assessed that China’s Ministry of State Security had recruited her to gather information on him, The New York Times reported.

China’s Foreign Ministry has called the claim unsubstantiated. Huang admitted to an affair on Nov. 8, and Nvidia’s board opened an independent investigation. Lawmakers have also called for congressional scrutiny.

Following this news, Bernstein cut Nvidia to Market-Perform from Outperform and halved its price target to $200. Analyst Stacy Rasgon said the stock is very cheap, but more allegations may surface and the choice of successor adds uncertainty. "On the positive, the fundamentals of the business (if not the leadership) are clearly still intact with numbers that remain massive (and probably still headed higher)," he wrote.

Oppenheimer also downgraded the stock to Perform and removed its price target. Analyst Rick Schafer warned of a potential "brain drain," naming AMD, Marvell and Broadcom as possible beneficiaries. "His departure presents a material, potentially thesis-changing event," he said.

Stifel kept its Buy rating but cut its target to $205 from $315, treating a leadership change as its base case. Citi kept Buy but lowered its target to $225, suggesting Kress could serve as interim CEO, as has now happened.

Mizuho, UBS and Raymond James maintained bullish ratings, with UBS flagging a possible pause in Nvidia’s funding of AI companies, which could temporarily remove about $50 billion from the ecosystem, along with the risk of tighter export restrictions.

Nvidia is due to report third-quarter results Nov. 17.

Apple stock is an AI "safe haven," Bernstein says

Bernstein has reiterated its Outperform rating and $380 price target on Apple, arguing the iPhone maker has become a safe haven for investors fleeing riskier AI stocks and that its record-high valuation is justified.

Analyst Mark Newman noted Apple spends just 2% of revenue on capex and has partnered with Google for its AI models, many of which run directly on devices rather than on expensive AI servers. That leaves Apple protected from both a pullback in data center spending and the shrinking free cash flow hyperscalers are experiencing.

"Furthermore, Apple could stand to benefit if/when the AI bubble actually bursts (e.g. from potentially cheap acquisition targets)," he wrote.

That status has pushed Apple’s valuation to about 37 times 2027 consensus earnings, an all-time high. Among large U.S. companies with expected earnings growth below 15%, only Apple, Costco, Walmart and GE trade above 30 times forward earnings.

Bernstein said Apple now trades in line with high-quality consumer stocks rather than tech peers, and looks cheaper than Costco and Walmart despite stronger revenue growth, cash generation and shareholder returns. Newman also compared Apple’s roughly 90% loyalty among iPhone users upgrading every four years with Costco’s 90% annual membership renewal rate.

"Is AAPL’s multiple outrageous? Considering its customer loyalty, earnings durability, growth, FCF generation and shareholder returns, we don’t think so," Newman said. "Our reverse discounted cash flow (DCF) implies a terminal growth rate of 4%, which appears reasonable."

Redburn cuts Broadcom to Neutral on limited re-rating potential

This week, Redburn downgraded Broadcom to Neutral and slashed its price target to $398 from $595, saying hurdles tied to its frontier AI customers make a re-rating unlikely to come quickly.

Analyst Mike Harrison noted Broadcom’s guidance implies it trades at about 11 times fiscal 2028 earnings, but that "counts for little when both the quality and the durability of earnings are under scrutiny."

He said Broadcom’s alignment with Anthropic and OpenAI means the labs "need to establish sustained primacy over open models and become self-financing in short order," while the market also needs reassurance that Google’s efforts to build its own tooling won’t become widespread. Clearing those hurdles could offer appreciable upside, but "is unlikely to occur quickly," he wrote.

Up to 15 gigawatts of compute for Anthropic and OpenAI underpins Broadcom’s $230 billion fiscal 2028 AI revenue guidance, but both labs rely heavily on credit market access, leaving the buildouts vulnerable to contagion from elsewhere in the AI ecosystem. Falling inference costs also threaten the labs’ dominance versus open models.

Broadcom’s custom silicon can give them a cost advantage, but competitive pressure will likely push inference gross margins lower. "Furthermore, the early traction Meta has made with its consumer-facing agentic AI platform implies that the locus of competition is evolving from model intelligence + harness, to harness + distribution," Harrison added.

Harrison also questioned the durability of the custom chip model, calling Google’s efforts to diversify away from Broadcom an important analogue. If the strategy works, the market is likely to assume frontier labs eventually follow.

Redburn also changed its valuation approach, saying its previous DCF model "implicitly endorses a rerating that is hard to underwrite at this juncture." "Strong earnings momentum is a better predictor of outperformance than low valuation multiples for tech companies," Harrison said.

JPMorgan upgrades SAP to Overweight amid AI shift to production

JPMorgan has upgraded SAP to Overweight from Neutral and raised its price target to €275 for December 2028 from €175 for December 2027, saying the shift of enterprise AI from pilots to production favors incumbent systems of record.

Analyst Toby Ogg said that as AI projects move into production with agents, "scale and return on investment (ROI) depend on much more than the model, namely context, control and cost." Prototyping agents is relatively simple, but production depends on data quality, process context and integration, which SAP supplies through its Business Data Cloud, Business Technology Platform, Signavio and Knowledge Graph, increasingly for non-SAP data too.

Ogg pointed to July’s Hugging Face breach, carried out end-to-end by autonomous agents, as a sign of the risks agents can pose. SAP’s Joule assistant runs within customers’ existing permissions and audit trails, and the company commits not to use customer data to train external large language models.

IDC found that 67% of organizations with AI agents in production have exceeded their budgets. "SAP can over time price AI on completed tasks, help route tasks to the most cost-effective model, and leverage business context for performance and cost advantages," Ogg said.

There are early signs in the numbers. SAP’s current cloud backlog has surprised to the upside for three quarters, with third-quarter growth of 26% implying a step up in absolute organic additions from a year earlier, "bucking the numerical 2nd derivative trend to the upside," Ogg wrote.

While SAP has recovered from its July lows and is "no longer priced for AI disruption," Ogg said the earnings growth algorithm alone supports attractive returns, with roughly 17% annual non-IFRS EPS growth through 2028. The target assumes about 24 times forward earnings.

KeyBanc lifts Palo Alto, says recent pullback offers attractive entry point

KeyBanc Capital Markets has upgraded Palo Alto Networks to Overweight from Sector Weight with a $380 price target, saying the stock’s sharp pullback over the past month creates an attractive entry point as cybersecurity spending is set to accelerate.

Analyst Eric Heath expects security budgets to grow materially in 2027 as AI drives a more dangerous threat environment and the market for securing AI matures. CISOs at a KeyBanc event in New York suggested budget growth would accelerate from single digits early this year to the mid-teens by year-end, and to more than 20% in 2027.

"Even among this CISO group expecting more moderate growth, they remarked how you should ’never let a good crisis go to waste,’ and they have an ’open checkbook,’" Heath wrote.

KeyBanc’s reseller survey also pointed to strength, with all partners meeting or beating plans in the third quarter and security growth accelerating for a fourth straight quarter.

Heath said Palo Alto, along with CrowdStrike, is a key beneficiary as customers consolidate security spending with fewer vendors, citing its broad lineup across securing AI, security operations, identity and network security.

"Further, we see the platform vendors with the richest volume of proprietary data best positioned to provide agentic cybersecurity capabilities and/or be defensible to incremental competition from LLM vendors expanding into the SOC," he said.