RBC Capital Markets now runs its US equity models on the bottom-up consensus minus 10%. Earlier in the summer the haircut was 5%.
The extra five points are not a forecast that AI earnings will disappoint. They are a margin of safety against the possibility, added while the rest of the market was still taking the consensus at face value.
"So look, one of the things we've been doing in our modeling is just taking the bottom-up consensus and haircutting it by 10%."
Lori Calvasina runs US equity strategy at RBC Capital Markets, publishes a weekly note that told clients last week to expect a 5% to 10% drawdown, and has moved the firm to a rolling 12-month price target instead of a year-end number.
The full segment is covered here so you can skip it.
Here are the 6 calls that matter.
👤 Guest: Lori Calvasina, Managing Director and Head of US Equity Strategy at RBC Capital Markets
🎙️ Host: Jonathan Ferro, a Bloomberg Surveillance anchor
📰 Published: 14 September 2026 on Bloomberg Surveillance
🔴 YouTube | 🟢 Spotify | 🟣 Apple Podcasts | ⏱️ length not available
Key Takeaways
RBC's models now take 10% off the bottom-up earnings consensus, double the haircut it used earlier in the summer
The first 5% covered the Iran war and inflation; the second 5% covers froth on the AI side
The capex story to watch is the 490 stocks that are not in the top 10 by market value
The mega-caps are at peak capex growth; the rest are, in her words, at about the third innings
She expects a 5% to 10% drawdown, and the reasons are four separate ones rather than AI alone
The historical break point in a hiking cycle is four hikes in 12 months, not the first one
RBC's rate strategist has moved from no hikes to three, and calls them adjustment hikes
A rolling 12-month target has replaced the December 31 number, because short-term damage gets paid for inside the year it happens
Last year's tariff drawdown ran 18% to 19% and the year still finished healthy
The consensus inflation forecast for the middle of next year assumes gasoline prices normalize, and the war threatens that assumption
1. A 10% Haircut On Earnings
The hour opened with stocks falling on the possibility of an AI pullback, after Anthropic's Dario Amodei laid out a three-step plan aimed at pacing the frontier. The host's first question to Calvasina was how much earnings risk is building.
The answer is a number RBC has already put into its models. "So look, one of the things we've been doing in our modeling is just taking the bottom-up consensus and haircutting it by 10%."
The size of the haircut has doubled since the early summer. The original 5% was there to cover the non-AI part of the index — pressure from the Iran war and from inflation. The extra 5% was added to account for froth on the AI side
She was careful not to present it as a call. She said RBC was not pounding the table that it would happen, but that concern had been building for a few months, and the firm wanted the numbers to reflect it and to calm investors' nerves
2. Capex: Look Past Top 10
Asked whether a slower build-out is coming, Calvasina pointed to a set of charts she said has become one of the more popular in RBC's deck.
S&P 500 capex growth hit new year-over-year highs in the second-quarter updates, which is not what a market worried about a capex pause would expect to see
The split inside the index is the whole point. The top 10 companies by market value — where the large AI spenders sit — have been at what she called peaks. The other 490 are at the start of their own cycle
Her instruction to clients is to change where they are looking. "We've been telling people, if you're concerned about the capex story going forward, take your attention off those top 10 market cap names." On that second group, "So you're up maybe third innings."
The example she gave came from insurers. "I got very excited reading insurance companies." She called a healthcare analyst and an insurance analyst about it after one company said it was not engaged in FOMO and then described the upgrades it was considering anyway, with AI prompting the thinking
What she is tracking is the second-order effect rather than the AI trade itself — capex and IT upgrades happening at ordinary companies, where AI may be one cause among several
3. Why She Sees A 5-10% Dip
The host asked how immune the derivative capex story is to the frontier models themselves, and to the coming listings of Anthropic and, next year, OpenAI.
She declined the question rather than answering it. "I'm not sure I have the answer to that question, to be honest." What she would say is that the market is full of jitters
The call itself was published before the week began. "And we made a comment last week in our weekly that we were expecting a 5% to 10% pulldown." RBC's view is that those risks have risen
The list of reasons is four items long and only one of them is AI. Midterm elections, where the last couple of cycles have produced extreme volatility in both directions in the back half of the year. Interest rate risk. A new Fed chair. And the war
On rates, her point is about correlation rather than level. "When the bond market and macro investors are worried about interest rates rising, everything else hits harder."
On the Fed, the precedent is the transition rather than the person. "You typically see volatility in markets when you're ushering in new leadership at the Fed." She said the same was visible under Kevin Warsh's three predecessors
On the war, her evidence is what companies have stopped saying. Two reporting seasons ago many of them expected it to be wrapped up by the summer or the second quarter. It was not, and this last reporting season they largely did not mention it
4. Four Hikes Is The Line
Pushed on why anyone should buy a 5% to 10% pullback with the war unresolved, diesel above $6 a gallon and rates pressuring multiples, Calvasina made the case in two parts.
The first is that no single factor sets the market. RBC's economists moved their GDP view up over the summer to a low 2% number, which she said is traditionally a very favorable environment for equities. "So I think we have to sort of get through this period and then look and see, do we still think the underlying economy is strong? Right now the numbers suggest yes."
The second is a count of hikes, and it is the sharpest number in the segment. "If it's a 12-month time frame and you're doing like one to four hikes, traditionally the market's actually been pretty strong." Past that, "But when you do more than four, right, if we start talking about six, seven, eight in a 12-month time frame, that's when the stock market tends to struggle."
RBC's own rate strategist has just moved. Blake Gwynn had not been looking for hikes. "He's changed his call now. He's looking for three." He describes them as adjustment hikes
That distinction is what keeps her constructive. An adjustment-hike world means short-term volatility that the market can recover from. A different story — more hikes, on her modeling — means bigger problems
5. Rolling 12-Month Targets
Asked what would have to happen to bring RBC's price target down, Calvasina explained why the question is harder to answer than it used to be.
The firm has moved to a rolling 12-month forward target and uses it as a longer-term bogey rather than adjusting a December 31 number in the short term
Short-term moves are handled through a separate framework she calls four tiers of fear, which is where the 5% to 10% drawdown estimate comes from
The reason is an observation from last year's tariff sell-off. The market fell something like 18% to 19% and the year was still a healthy one
What she found striking was the investor behavior, not the index level. Out on the road in June, clients told her they were done with 2025 and looking ahead to 2026, on the reasoning that whatever damage tariffs did had already been paid for. "So that's one thing that I think makes price targets very tricky when you're trying to sort through short-term issues."
6. Gas Vs The 2% Forecast
The host put gasoline at $4.30 a gallon nationally and diesel above $6, and asked how hard that lands on smaller companies.
Her focus is the consumer and the inflation forecast rather than small-cap earnings. Consensus has headline inflation running around 2% and change year over year by the middle of next year
That forecast contains an assumption she does not think is being tested. Talking to economists across the street, she said those numbers appear to bake in gasoline prices normalizing — and the war threatens exactly that
Nothing has moved yet. She said the economics community's numbers have not shifted, and flagged it as something to pay attention to rather than as a call
Bonus Insights
RBC has not yet produced a 2027 year-ahead forecast. Asked whether she is already thinking about 2028, Calvasina said you have to think earlier and earlier, then added: "I mean, we haven't done a 2027 year ahead forecast yet."
The switch to a 12-month forward target has had a side benefit she did not expect. She said it forces the team to separate short-term issues from long-term ones, which is why the framework has held up through this year
RBC updates all of its models once a month, and Calvasina said the longer-term picture is being reassessed constantly. "But right now, we're not seeing the longer-term story derailed."
Calvasina's bottom line is that the risk to equities from here is a 5% to 10% drawdown driven by four things at once, not by AI alone, and that the position to hold through it is a constructive one as long as the economy grows at a low 2% pace and the Fed stops at three or four hikes.
Products, Companies & Tools Mentioned
RBC Capital Markets (Calvasina's firm: its equity strategy team now haircuts the bottom-up consensus by 10%, and its rate strategist Blake Gwynn has moved from no hikes to three)
Anthropic and OpenAI (Anthropic's chief executive set off the session with a plan to pace the frontier; the host asked whether the wider capex story is immune to the two companies' coming listings, and Calvasina said she does not have the answer)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: