← Analysis page  ·  CNBC Halftime Report hub  ·  Research hub

Actionable insights — AI Safety Concerns Push Tech Stocks and the Market Lower

The repeatable analysis behind the committee's calls: not what they bought, but how they got there — written so the process can be rerun later on different names.
2026-SEP-14 · CNBC Halftime Report (audio edition) · Santoli + committee (Terranova, Thomas, Weiss, Lebenthal) + Kate Rooney · ▶ Listen · full analysis · transcript
How to read this page: the recurring move in this episode is sorting a shock by who pays and who saves — a slower AI frontier costs the vendors and the debt-funded industrials, and saves the spenders; a hike into a supply shock costs Main Street and does not fix the supply. Each insight below is written as a procedure you can rerun on names this episode never mentions; the boxed Here: line shows how it played out on 2026-SEP-14.

1. When a capex theme is threatened, split the chain into spenders and vendors before you sell anything

The repeatable method
  1. Map the theme's money flow: who writes the cheques (spenders), who receives them (vendors), and who is two steps out living on multi-year order books.
  2. For a "slowdown" headline, ask which side's earnings rise and which fall. Less spending is a margin gain for a spender that already monetizes the theme.
  3. Check the tape for confirmation that the market is drawing the same line — spenders green, vendors red on the same day.
  4. Rank vendors by how far out their value sits. The further the value depends on years 3–5 of the order book, the bigger the hit from an uncertain future.
  5. Hold or add to the spender side; reduce or refuse new money on the vendor side until the headline resolves.
Here: Santoli: "maybe the spenders are going to be OK, even if the vendors have to do with less in the future." The tape agrees — SOX −5.3% while MSFT and GOOGL are green; Weiss: "Meta and Microsoft and Alphabet move up because say, OK, great, they're going to slow down spending." The furthest-out vendors take the worst of it: AI-power industrials "getting smoked because that's the area where it really matters what's going to happen in 3, 4, 5 years."
Watch for

2. Ask who funds the backlog — then watch the debt market, not the order book

The repeatable method
  1. For any company whose bull case is a record backlog, identify the customers' funding source: cash flow, equity raises or debt issuance.
  2. If it is debt, the backlog is only as good as the issuance window. Two things close that window: slower growth expectations and higher yields.
  3. When both arrive together, treat the backlog as contingent even though nothing has been cancelled.
  4. Track the leading indicator — new issuance by the end customers — rather than waiting for order cancellations, which come last.
Here: Terranova on his JOET AI-power names: "if you tell me we're slowing the rate of spending and if you tell me that yields are going to remain high, that's a disincentive to see more supply in the debt market… that is absolutely going to affect these industrial names… they have thrived over the last eight quarters on the continual spend and the build." Santoli supplies the funding gap: "they're living off of the forward order books for which nobody has the money to pay for." Weiss adds the private-market version: AI "lives and breathes" in private markets, so unprofitable labs must raise more — "a thinning of the herd."
Watch for

3. Don't buy what is near a 52-week low — buy what is at a 52-week high, and write the rule down

The repeatable method
  1. Score the stock's momentum on at least two horizons (12-month and 3-month).
  2. If both are negative and the price is near a 52-week low, do not buy — regardless of how good the forward fundamentals look.
  3. Define "prove itself": the decline stops, a base forms, and the recovery begins. Enter then, accepting you miss the bottom.
  4. Apply the same rule to names you already own that break down — let the scheduled rebalance make the decision rather than improvising.
  5. Look for the mirror image for new money: names at 52-week highs on a down day.
Here: Terranova on APP after Morgan Stanley's cut to $450: "This is a broken momentum stock… 42% on a 12 month momentum score… three month… down as well… I don't like buying stocks that are near a 52 week low despite what might be the vision of some really good fundamentals ahead… let it prove itself first." His final trade is the mirror: EXPD "hitting a 52 week high today as we speak" on a red tape. In the airlines, the same test splits the pair — DAL above its 200-day, "United's broken down."
Watch for

4. Let the stock's reaction to good news grade the thesis

The repeatable method
  1. After a report, record the price reaction alongside the numbers. A company that "reported good numbers" and cannot bounce is telling you the market's marginal buyer is gone.
  2. Two consecutive reports with no durable bounce is a pattern, not noise.
  3. If you tested it with a position and got out flat, do not re-enter lower just because it is cheaper — the price below your exit is the verdict.
  4. Combine with valuation: a broken reaction on a still-high multiple is the falling-knife setup.
Here: Weiss on AVGO: "they missed arguably this time, they missed last time… they reported good numbers, guidance, squishy things in their margins. The stock has never bounced… I bought it after the prior quarter, bounced a little bit, sold it flat and it traded at 410. Now it's back down to… 346. To me, that's the message in the market. You can pick a falling knife here, particularly when you still have valuations that are too high."
Watch for

5. Before cheering a hike, classify the inflation — demand or supply

The repeatable method
  1. Decompose the inflation driving the policy move: is it excess demand (hikes work) or a supply shock (hikes cannot open a strait or build a refinery)?
  2. If supply, expect the hike to hit the rate-sensitive economy (housing, small business) while leaving the inflation source untouched — the stagflation setup.
  3. Check real growth, not nominal: 6% nominal on 1½–2½% real is not an overheating economy.
  4. Pick the historical analogue deliberately. If the central bank's target is far away and it has committed publicly, a single move (1997) is less likely than a cycle (1999).
  5. Look for the pass-through that hasn't happened yet (diesel into food and freight) — it decides whether the next meeting is live.
Here: Weiss: "tightening to reel in a supply shock doesn't work. And what we're seeing with oil is supply shock"; Lebenthal: "They can't open the Strait of Hormuz… they are probably backed into raising rates more than once." Thomas runs the classification to its end: the 10-year hit 5% on supply, "not because the economy is overheating"; a hike is "a reaction to the market, not to the economy"; and the analogue is "more like 1999, because Kevin Warsh has committed so much to that 2% inflation target. 25 basis points isn't going to get us there." Lebenthal's pass-through: energy spikes "haven't yet flowed through"; diesel at a record, no new capacity.
Watch for

6. When the index is near its highs on a pending resolution, measure how many exits there are

The repeatable method
  1. Ask why the market is only a few percent from its high despite visible headwinds. Name the assumption holding it up.
  2. If the assumption is time-bound ("this resolves in 60–90 days"), positioning is likely one-sided.
  3. Estimate the exit: if the assumption fails, who sells and into what liquidity?
  4. You don't have to act on it — but know it is there, and size and hedge accordingly.
  5. Cross-check breadth: a cap-weighted index holding while the equal-weight breaks its 50-day is the same fragility from another angle.
Here: Terranova: "The risk actually is… that the market is still 3% from the high… It is not because people believe 60 to 90 days from now this all resolves itself… everyone is positioned one way. And if they are not receiving what their expectation was, they are very quickly exiting from what will only be 1 door. I don't subscribe to it… but you always have to know it's there." Santoli's breadth version: the S&P is "taking it like a champ" while equal-weight "broke the 50 day moving average"; Thomas: "concentrated large cap leadership is a more fragile place to be."
Watch for

Methods distilled from the public CNBC Halftime Report audio episode of 2026-SEP-14 (transcript in transcript.txt). Not investment advice. © CNBC for source material.