This website uses cookies

Read our Privacy policy and Terms of use for more information.

The AI Investing Pulse

September 17th, 2026

In this Week’s Edition:

  1. Analysis - The AI Warning Sent Money Into Cybersecurity. These 3 Stocks Score Highest

  2. Stock Ideas - Palantir, Marvell and Snowflake Named as Pullback Picks

  3. News - The Easy Phase of the AI Trade Is Over

  4. Startups - Samsung Co-Leads a $230m Round in Chip Startup Euclyd

  5. Trends - Could AI Wipe Out Humanity Within a Decade? Experts Split

  6. Other News - Musk Says Data Centres Cut Power Bills. The Data Disagrees

The AI Warning Sent Money Into Cybersecurity. These 3 Stocks Score Highest

The people building AI spent the weekend warning about it.

It started the week before, with a resignation. Jacob Coxon, a researcher at Anthropic, quit and wrote on X that neither Anthropic nor OpenAI was acting responsibly. A second Anthropic researcher, Evan Hubinger, said he puts the chance of AI killing all humans within the next ten years at more than 10%, according to Benzinga. Geoffrey Hinton, often called the godfather of AI, has said a 10% chance over a decade "seems not an unreasonable estimate".

On Saturday 13 September, Anthropic's chief executive Dario Amodei published an essay titled "We Must Pace the Frontier". He argued that AI companies should slow the pace at which they improve their models. Sam Altman of OpenAI replied "I agree with Dario". Elon Musk wrote "Dario is right". Altman also told Fortune that OpenAI will not go public this year, citing safety.

On Monday the Philadelphia Semiconductor Index fell almost 6%, its worst day since July, NBC News reported. President Trump called fears of AI destroying humanity a "HOAX" and said AI "will not be stopped" while he is in office. On Wednesday the Federal Reserve raised interest rates by a quarter of a point, its first rise since 2023.

Where the money went

We sort every company in our universe into one of five layers, based on what it does in the AI chain. Think of it as a building site. Power is the electricity supply. Build is the people making the tools and materials. Connect is the wiring. Develop is the architects writing the software. Deploy is everyone using the finished building. Two hundred companies, five jobs.

Layer

This week

Past 3 months

Universe average

+0.1%

+0.6%

Develop

+3.5%

+15.4%

Deploy

+3.0%

+17.3%

Connect

-2.7%

-11.1%

Build

-3.3%

-19.4%

Power

-4.0%

-21.5%

Across all 200 companies, the week was flat. Underneath, money moved from the hardware layers to the software layers, and the three-month column shows that move was already under way.

The sharpest part of it was cybersecurity. Half of the ten biggest risers in our universe were security companies. CNBC reported the reason: security companies are paid to defend against AI-related attacks, and investors expect more spending on security, not less. A warning about AI risk is, for them, a sales argument.

Three security companies that score highest on our test

This is not a buy list. We ran our scoring test across the security companies in our universe, and these three came out on top. The reason is the same for all three. Each grew revenue by about a quarter on the year in its latest quarter, keeps around three-quarters or more of every sales dollar as gross profit, and generates real cash. The rally did not pick them, and a strong score is not a verdict on the business.

Each one protects a different part of a company.

1. Fortinet, Inc. (FTNT)

  • Sector / Industry: Information Technology | Software

  • AI Stack Layer: Deploy, Connect

  • Market Cap: $126.0bn

  • 1 Week: +8.1% | YTD: +116.3%

  • Fundamental Score: 88/100 | Momentum (RS): 90.0 (+3.5 on the week)

  • AIIP Ranking: Watchlist

Fortinet is best known for guarding the front door. Its FortiGate firewalls sit at the edge of a company's network and decide what gets in. It also sells networking equipment, device protection and security for the systems that run factories and infrastructure. If attackers start using AI to probe networks faster, that edge is the first thing they hit.

It is the strongest grower of the three on our measures. Revenue rose 26% on the year in its latest quarter, earnings per share rose 44%, and it keeps about 80 cents of every dollar of sales as gross profit. It rose the least of the three last week, at 8%, but it was already on our watchlist and has more than doubled this year.

  • Sector / Industry: Information Technology | Software

  • AI Stack Layer: Deploy, Connect

  • Market Cap: $31.2bn

  • 1 Week: +17.2% | YTD: -14.8%

  • Fundamental Score: 85/100 | Momentum (RS): 25.6 (+24.9 on the week)

  • AIIP Ranking: Fundamental

Zscaler checks every connection. Its main business is cloud security: instead of letting staff onto the whole network once they log in, it checks each connection between a person and an app, through its own cloud. If an attacker steals one login, that limits how far they can get. The company now applies the same approach to AI agents and AI services, not just people, which puts it close to this week's story.

This is the one the market had walked away from. Even after last week, the shares are down 15% this year and 31% over twelve months, while revenue grew 25% on the year last quarter. A week ago it sat near the bottom of our momentum ranking, and it climbed 25 points in seven days. The recovery started before the essay, though: the shares are up 54% over three months.

  • Sector / Industry: Information Technology | Software

  • AI Stack Layer: Deploy, Connect

  • Market Cap: $247.7bn

  • 1 Week: +15.6% | YTD: +106.0%

  • Fundamental Score: 79/100 | Momentum (RS): 91.7 (+6.0 on the week)

  • AIIP Ranking: Watchlist

CrowdStrike is best known for guarding each device. Its Falcon software watches laptops, servers and cloud systems for suspicious behaviour, and it is built to catch an attack that has already got past the front door. It also protects user identities and data, and sells services to secure companies' own generative AI systems.

And it has a direct link to the AI labs. CNBC reported that CrowdStrike and Palo Alto Networks have taken part in early testing of unreleased OpenAI and Anthropic models. Revenue grew 26% on the year last quarter and earnings per share more than doubled. It is not overlooked: it is on our watchlist, has roughly doubled this year and rose 27% over the past four weeks.

What the screen rejected

Not every security company in the rally scores well. Three that rose sharply fall short on our test.

  • Rapid7 rose 19%, the most of any security company we track. Its revenue shrank slightly on the year last quarter.

  • SentinelOne rose 17%. Revenue grew 21%, but earnings per share fell on the quarter and its gross margin slipped.

  • Palo Alto Networks, the largest security company in our universe, rose 11%. Revenue grew strongly, but quarterly earnings per share fell sharply and its gross margin slipped too.

What to be careful of

One week is a short window, and ours includes two trading days before the essay. Fortinet and CrowdStrike have already doubled this year, and Zscaler is still down for the year. Our scores read reported numbers only, not competition or management. Treat this as a starting point for your own research. It cannot tell you whether any of these companies suits your circumstances, and if you want help with that, it belongs with a regulated financial adviser. Past performance is no guide to what comes next.

Our take

This week the people building AI said it needs to slow down. The US and Chinese governments both said no. President Trump called the fears a hoax, and China's foreign ministry dismissed them as "fearmongering", NBC News reported. Whether anyone actually acts on the warning remains to be seen.

What did change is the conversation. Whether or not AI slows down, rising risk makes a stronger case for spending on defence. That is the argument money followed into security this week, and we expect it to come up more often as AI gets more capable.

The test is in the numbers. A rally on a news cycle is not the same as orders, and the next quarterly reports from these three will show whether growth holds.

None of which tells you what to do with it. This is where the money went this week, not where it goes next.

Questions readers ask

Why did cybersecurity stocks rise after the AI warnings? CNBC reported that security companies are paid to defend against AI-related attacks, so warnings about AI risk point to more spending on their products. Half of the ten biggest risers in our 200-company universe over the five trading days to 16 September were security companies.

Which cybersecurity companies score highest on the AIIP test? Fortinet, Zscaler and CrowdStrike score highest of the eleven security companies we track, on revenue and earnings growth, gross margin and free cash flow.

Did the stocks that rose most also score best? No. Rapid7 and SentinelOne both rose 17% or more but score poorly on our test, as does Palo Alto Networks.

Is this a recommendation to buy these three? No. AI Investing Pulse publishes research, not advice. These three are the output of a mechanical screen, shown with the reasoning attached so you can disagree with it.

IMPORTANT LEGAL DISCLAIMER

Not Investment Advice: This content is provided by AI Investing Pulse for informational and educational purposes only. It does not constitute investment advice, a personal recommendation, or an invitation or inducement to engage in any investment activity.

Not Regulated: AI Investing Pulse is not authorised or regulated by the Financial Conduct Authority (FCA) in the United Kingdom, is not registered with the Securities and Exchange Commission (SEC) or any state securities regulator in the United States, and is not registered with the Canadian Investment Regulatory Organization (CIRO) or any provincial securities commission in Canada.

Methodology Disclosure: The AIIP Index scores and rankings mentioned in this article are generated by a proprietary quantitative methodology based on publicly available financial data. Our full methodology is explained in the "About AIIP" section below. These scores are objective system outputs, not recommendations or endorsements.

Risk Warning: Investing in stocks involves risk, including the potential loss of principal. Past performance of stocks, scores, or rankings is not indicative of future results. Stock prices can decline as well as rise, and you may lose some or all of your invested capital.

Third-Party References: References to analyst opinions, bank research, media publications, or the term "picks" refer to third-party selections, not AIIP recommendations. We aggregate this information for educational analysis only.

Affiliate Links: Ticker symbols in this article link to StockAnalysis.com through an affiliate arrangement. If you take out a paid StockAnalysis plan through one of those links, AI Investing Pulse may receive a commission at no additional cost to you. The links are applied mechanically to every company we cover and play no part in which companies appear, how they are scored, or how they are ranked.

Seek Professional Advice: Always consult a qualified, regulated financial professional who understands your personal circumstances before making any investment decisions. Consider your individual financial situation, risk tolerance, investment objectives, and time horizon.

About AIIP - The AIIP Index tracks 200 AI-focused public companies across the full AI stack, serving as our benchmark for sector performance. All scores are proprietary and calculated using data from Finbox (powered by S&P Global Intelligence).

AIIP Fundamental Score (0–100) combines sales and EPS growth with financial quality, which is measured on gross margin, the direction of that margin, and free cash flow. Valuation is applied as a threshold rather than a weighting: companies with a forward PEG above 5.0 are excluded.

AIIP Relative Strength (RS) Score (0–100) measures a stock's price performance against the other AI stocks in the index. It combines twelve-month and six-month performance and applies a penalty for trading well below a 52-week high. It is a measure of price, not of business quality.

Combined Score (0–200) is the Fundamental Score plus the RS Score. It is shown for reference and does not itself determine ranking.

Ranking Status is set by the two scores independently: Fundamental: Fundamental Score ≥ 70, RS < 80. Momentum: RS ≥ 80, Fundamental Score < 70. Watchlist: Fundamental Score ≥ 70 and RS ≥ 80. No Signal: both below those thresholds.

TOP AI STOCKS PERFORMANCE

COMPANY

SECTOR

WEEKLY

Schrodinger (SDGR)

Healthcare

27.3%

Veritone (VERI)

Technology

23.8%

Tempus (TEM)

Healthcare

19.1%

TOP AI ETFs PERFORMANCE

ETFs

SECTOR

WEEKLY

Robo Global (THNQ)

Global AI

0.7%

Global X (AIQ)

AI & Tech

-0.1%

KraneShares (AGIX)

AI & Tech

-0.8%

AI Stock Ideas

Three AI names picked in case September's seasonal weakness bites. Palantir grew revenue 93% last quarter, and Marvell's multi-year custom-chip deal with Alphabet is the core of its case.

All three raised guidance in their latest results. Broadcom's AI semiconductor revenue rose 221% to $16.7bn in fiscal Q3, and Marvell's investor day on 6 October is the next marker.

CEO Michael Intrator says the company is "struggling to meet demand every day". Q2 sales rose 112% to $2.6bn, but $640m of interest costs still kept it loss-making.

Nvidia, TSMC, Arista, Dell and HPE are picked as names to consider if Monday's sell-off overshoots. Arista's Q2 revenue rose nearly 38%, and Dell's AI-server revenue doubled to $16.4bn.

AI Stocks & ETFs News

The tech sector's forward P/E has fallen from 29 to 21 in a year, and chip stocks were already 20% off their June highs before Monday's slide. The silver lining: the froth has largely been purged.

Reuters reports the listing could raise up to $100bn at a valuation as high as $2.3tn, targeted before the November midterms. Talks are ongoing and nothing is final.

The Amazon agreement could bring up to $60bn in payments over a decade. Qualcomm targets $5bn of data-centre revenue by fiscal 2027 and more than $15bn by 2029, with Amazon orders due from the December quarter.

A screen of eight indicator categories finds most near levels that preceded past peaks, and the firm forecasts an S&P 500 fall of at least 30% starting next year. Morgan Stanley's Lisa Shalett disagrees and holds an 8,000 year-end target.

The Philadelphia Semiconductor Index fell 5.8% on Monday after Dario Amodei called for slower AI development, with Arm down 9.7% and Nvidia 3.4%. Sam Altman also said OpenAI will not go public this year, citing safety.

S&P says hyperscaler credit quality is "gradually weakening" as they lend their ratings to neoclouds, data-centre operators and AI labs. It expects the top six to spend more than $7tn on AI capex through 2030.

AI Startups

The Dutch company is designing a non-GPU system for AI inference, though it is unproven at scale and does not plan to ship until 2028. Samsung brings memory and supply-chain know-how as well as cash.

The Cisco and Arista rival launched a new network architecture and a Qualcomm partnership alongside the round. Its pitch is simple: slow networks leave expensive AI chips sitting idle.

Anthropic researcher Evan Hubinger puts the odds of AI killing all humans in the next ten years above 10%, and Geoffrey Hinton calls that "not an unreasonable estimate". Gary Marcus sees no realistic path to it.

"Whoever wins AI, wins," Trump said, dismissing calls to pace frontier development after Nvidia fell 3% and AMD 4.4%. Without regulation or a deal with China, voluntary restraint by the labs looks fragile.

Others

US electricity prices rose 4.1% in the year to August, ahead of 3.4% headline inflation. The $180 annual saving Musk cited applies to one Georgia utility's contracts.

How did you like this week's newsletter?

Click to vote, we'll use your feedback to improve the next one.

Login or Subscribe to participate

Not Investment Advice: This content is provided by AI Investing Pulse for informational and educational purposes only. It does not constitute investment advice, a personal recommendation, or an invitation or inducement to engage in any investment activity.

Not Regulated: AI Investing Pulse is not authorised or regulated by the Financial Conduct Authority (FCA) in the United Kingdom, is not registered with the Securities and Exchange Commission (SEC) or any state securities regulator in the United States, and is not registered with the Canadian Investment Regulatory Organization (CIRO) or any provincial securities commission in Canada.

Methodology Disclosure: The AIIP Index scores and rankings mentioned in this article are generated by a proprietary quantitative methodology based on publicly available financial data. Our full methodology is explained in the "About AIIP" section below. These scores are objective system outputs, not recommendations or endorsements.

Risk Warning: Investing in stocks involves risk, including the potential loss of principal. Past performance of stocks, scores, or rankings is not indicative of future results. Stock prices can decline as well as rise, and you may lose some or all of your invested capital.

Third-Party References: References to analyst opinions, bank research, media publications, or the term "picks" refer to third-party selections, not AIIP recommendations. We aggregate this information for educational analysis only.

Affiliate Links: Ticker symbols in this article link to StockAnalysis.com through an affiliate arrangement. If you take out a paid StockAnalysis plan through one of those links, AI Investing Pulse may receive a commission at no additional cost to you. The links are applied mechanically to every company we cover and play no part in which companies appear, how they are scored, or how they are ranked.

Seek Professional Advice: Always consult a qualified, regulated financial professional who understands your personal circumstances before making any investment decisions. Consider your individual financial situation, risk tolerance, investment objectives, and time horizon.

About AIIP - The AIIP Index tracks 200 AI-focused public companies across the full AI stack, serving as our benchmark for sector performance. All scores are proprietary and calculated using data from Finbox (powered by S&P Global Intelligence).

AIIP Fundamental Score (0–100) combines sales and EPS growth with financial quality, which is measured on gross margin, the direction of that margin, and free cash flow. Valuation is applied as a threshold rather than a weighting: companies with a forward PEG above 5.0 are excluded.

AIIP Relative Strength (RS) Score (0–100) measures a stock's price performance against the other AI stocks in the index. It combines twelve-month and six-month performance and applies a penalty for trading well below a 52-week high. It is a measure of price, not of business quality.

Combined Score (0–200) is the Fundamental Score plus the RS Score. It is shown for reference and does not itself determine ranking.

Ranking Status is set by the two scores independently: Fundamental: Fundamental Score ≥ 70, RS < 80. Momentum: RS ≥ 80, Fundamental Score < 70. Watchlist: Fundamental Score ≥ 70 and RS ≥ 80. No Signal: both below those thresholds.