캐피털 원(COF) 2026년 3분기 실적 콘퍼런스 콜: NIM 상승, 디스커버 시너지 진전

캐피털 원 파이낸셜(Capital One Financial Corporation, NYSE: COF)이 2026년 3분기 순이익 35억 달러(희석 주당순이익 5.61달러)를 기록했다고 발표했다. 디스커버 및 브렉스 인수 관련 조정 항목을 제외한 조정 주당순이익(EPS)은 6.69달러다.

매출은 전분기 대비 3% 증가한 반면 비이자비용은 2% 감소하며 대손충당금 적립 전 영업이익이 10% 증가했다. 경영진은 마케팅 및 기술 분야에 대한 투자 확대가 회사 전략의 핵심으로 유지될 것이지만, 통합 시너지 측면에서 디스커버 통합 작업은 예정보다 빠르게 진행되고 있다고 밝혔다.

핵심 요약

• 캐피털 원은 35억 달러(희석 주당 5.61달러)의 순이익을 기록했다. 디스커버 및 브렉스 관련 항목을 제외한 조정 EPS는 6.69달러다.
• 전분기 대비 매출 3% 증가와 2%의 비이자비용 감소에 힘입어 충당금 적립 전 영업이익이 10% 증가(조정 기준 9% 증가)했다.
• 순이자마진(NIM)은 국내 카드 수익률 상승, 평균 현금 잔액 감소, 분기 영업일수 1일 증가에 힘입어 전분기 대비 40bp 상승한 8.41%를 기록했다.
• 신용손실충당금은 2% 감소한 29억 달러를 기록했으며, 여기에는 31억 달러의 순대손상각과 2억 2,500만 달러의 충당금 환입이 포함되어 있다.
• 국내 카드 구매액은 전년 동기 대비 11% 증가했으며, 대손상각률은 3.98%로 하락했다. 디스커버를 제외한 기말 카드 대출은 약 6% 증가했다.
• 캐피털 원의 연초 대비 누적 실적에는 약 18억 달러 규모의 디스커버 통합 연간 시너지가 반영되었다. 경영진은 이러한 시너지가 예정보다 조기에 달성되고 있다고 밝혔다.

핵심 재무 데이터

사업 및 영업 실적

국내 카드

국내 카드 구매액은 전년 동기 대비 11% 증가했으며, 이는 주로 기존 캐피털 원의 성장에 기인한다. 브렉스 및 법인 카드 사업도 소폭 기여한 반면, 기존 디스커버 구매액은 일시적인 디스커버 대출 성장 둔화로 인해 거의 변동이 없었다.

기말 국내 카드 대출은 전년 동기 대비 약 2% 증가했다. 기존 디스커버 카드 대출은 약 3% 축소된 반면, 디스커버를 제외한 기말 대출은 약 6% 증가했다. 매출은 5% 증가했으며, 매출 마진은 17.8%를 기록했다.

국내 카드 대손상각률은 3.98%로 전분기 대비 73bp, 전년 동기 대비 65bp 하락했다. 연체율은 통상적인 계절성 요인으로 인해 2분기 대비 35bp 상승했으나 전년 동기 대비로는 15bp 하락한 3.74%를 기록했다.

비이자비용은 전년 동기 대비 9% 증가했으며, 이는 기술 투자, 마케팅 비용 증가 및 브렉스 편입을 반영한다. 전사 마케팅 비용은 14% 증가한 약 16억 달러에 달했다.

소비자 금융 및 결제 네트워크

글로벌 결제 네트워크 거래액은 약 1,910억 달러로 전년 동기 대비 25% 증가했으며, 이는 주로 캐피털 원 직불카드 고객의 디스커버 네트워크 전환을 반영한다.

자동차 대출 신규 발급액은 22% 증가했으며, 소비자 금융 기말 대출은 전년 동기 대비 약 110억 달러(13%) 증가했다. 매출은 주로 디스커버 매출 시너지와 자동차 대출 성장에 힘입어 약 17% 증가했다.

자동차 대출 대손상각률은 1.59%로 전년 동기 대비 5bp, 전분기 대비 16bp 상승했다. 경영진은 전년 대비 상승 원인을 주로 서브프라임 신규 대출 비중이 팬데믹 이전 수준으로 서서히 복귀하고 있기 때문으로 보았다. 자동차 대출 연체율은 전년 동기 대비 35bp 하락한 4.64%를 기록했다.

기업 금융

기말 및 평균 대출은 예금과 마찬가지로 전분기 대비 약 1% 증가했다. 매출은 비이자이익에 힘입어 2분기 대비 6% 증가한 반면, 비이자비용은 약 1% 증가했다.

요주의 정상대출 비율은 4bp 하락한 4.4%를 기록했고, 요주의 부실대출 비율은 10bp 하락한 1.22%를 기록했다.

디스커버 및 브렉스 통합

캐피털 원은 디스커버 통합 작업이 순조롭게 진행되고 있으며 통합 시너지가 예정보다 빠르게 실현되고 있다고 밝혔다. 회사는 직불카드 사업의 디스커버 네트워크 전환을 완료했으며, 캐피털 원 신용카드의 해당 네트워크 신규 발급 및 이관 작업을 테스트하고 있다.

9월 말 기준 디스커버 카드 신규 발급의 100%가 캐피털 원의 기술 스택에서 작동하고 있다. 경영진은 기존 디스커버 카드 포트폴리오의 전환이 1분기 내에 완료될 것으로 예상하고 있다.

브렉스의 경우 캐피털 원은 우선 규제 대응 및 리스크 관리 통합에 주력하는 한편, 해당 사업이 성장에 집중할 수 있도록 지원하고 있다. 장기적 계획으로는 브렉스의 지출 관리 도구를 캐피털 원 카드 상품, 여행 서비스, 은행 기능 및 결제 인프라와 연계하는 방안이 포함된다.

경영진 가이던스

• 경영진은 주로 계절적 요인으로 인해 3분기 대비 4분기 마케팅 비용 증가율이 2025년에 기록한 3분기 대비 4분기 증가율과 유사할 것으로 예상하고 있다.
• 캐피털 원은 2027년 동안 디스커버, 프리미엄 카드 고객, 캐피털 원 쇼핑 및 전국 단위 은행 사업을 포함한 마케팅 분야에 더 대규모로 투자할 계획이다.
• 디스커버 카드 대출은 플러스 성장으로 돌아서기 전까지 향후 수 분기 동안 계속 축소될 것으로 예상된다. 경영진은 인수 완료 이후 3분기에 처음으로 신규 대출 발급이 전년 동기 대비 증가했다고 언급했다.
• 경영진은 합의 추정치에 불확실성의 범위가 크지만, 남은 디스커버 시너지 효과와 추진 중인 비자 및 마스터카드 합의에 따른 예상 정산 수수료 매출 영향이 규모 면에서 거의 상쇄된다고 밝혔다.
• 캐피털 원은 향후 순이자마진(NIM) 변동이 주로 계절성과 자산·부채 구성 비율에 따라 결정될 것으로 예상하고 있다. 경영진은 디스커버 인수로 인한 구조적 NIM 효과가 대부분 일단락되었다고 설명했다.
• 회사 측은 기반 기술, AI, 디스커버 네트워크, 프리미엄 카드 고객 및 전국 단위 은행에 대한 투자를 지속하는 한편, 이러한 우선순위 영역 이외에서는 효율화를 추진할 계획이다.

리스크 및 주요 관찰 사항

• 캐피털 원은 국내 카드 대손충당금을 설정할 때 지정학적 리스크와 관련된 하방 경제 시나리오에 더 큰 가중치를 두었다.
• 비자 및 마스터카드와의 잠재적 합의로 인해 정산 수수료 매출이 감소할 수 있다. 경영진은 최종 승인 가능성이 상당한 것으로 보고 있으나 영향의 규모와 관련해 상당한 불확실성이 존재함을 인정했다.
• 디스커버 대출 잔액은 이전에 진행된 신규 발급 축소, 신용 한도 관리 및 고잔액 리볼빙 이용자에 대한 노출 축소의 여파로 여전히 압박을 받고 있다.
• 성장의 중심이 프리미엄 카드, 전국 단위 은행, 캐피털 원 쇼핑 및 기타 마케팅 집약적 사업으로 이동함에 따라 마케팅이 회사 비용 기반에서 차지하는 비중이 확대되고 있다.
• 신규 대출 중 서브프라임 비중의 점진적 정상화로 인해 자동차 신용 비용이 영향을 받을 수 있다.
• 캐피털 원의 공시 순이익은 2026년 중 충당금 환입의 수혜를 입었으며, 경영진은 기초 이익 창출 능력을 평가할 때 이를 정상화하여 고려해야 한다고 밝혔다.

애널리스트 Q&A 주요 내용

이익 창출 능력 및 투자: 경영진은 충당금 환입, 4분기 마케팅 비용, 남은 거래 효과 등을 조정한 후 디스커버 거래에서 기대되는 이익 창출 능력이 이미 가시화되고 있다고 밝혔다. 캐피털 원은 특정 수익률이나 효율성 비율 목표를 제시하지 않았으며, 투자 수준이 계속 높은 상태를 유지할 것임을 강조했다.

신용 실적: 경영진은 카드 대손상각률 개선의 원인을 이전의 연체율 개선 및 견조한 회수 실적 덕분으로 돌렸다. 최근의 연체율 추세는 상회세를 지속하기보다는 통상적인 계절성에 가까워졌다.

디스커버 네트워크 전환: 직불카드 전환은 완료되었으나, 신용카드 전환은 단일 대규모 전환이 아닌 점진적으로 진행될 예정이다. 캐피털 원은 네트워크 가맹점 수용성과 디스커버 브랜드에 투자하는 동시에 어떤 상품과 고객을 이관할지 테스트하고 있다.

AI 전략: 캐피털 원은 AI의 재무적 이점을 수치화할 계획이 없다. 경영진은 AI를 고객 경험, 소프트웨어 개발, 언더라이팅, 마케팅, 사기 방지, 고객 서비스, 리스크 관리 및 사이버 보안에 영향을 미칠 것으로 기대되는 포괄적 역량으로 설명했다.

대차대조표 복원력: 경영진은 보수적인 언더라이팅, 자본 관리, 저위험 지출 고객 비중 확대 및 실시간 모니터링 개선을 강조했다. 또한 현재 예상신용손실(CECL) 모형 및 재무상태표상 대출 구조가 글로벌 금융위기 당시 경험했던 것과는 다른 충당금 변동 양상을 만들어낼 수 있다고 언급했다.

실적 발표 컨퍼런스 콜 전문

---

전체 실적 발표 컨퍼런스 콜 녹취록

경영진 발표

Operator

Good day, and thank you for standing by. Welcome to the Capital One Q3 2026 Earnings Call. Please be advised that today's conference is being recorded. [Operator Instructions]

I would now like to hand the conference over to your speaker today, Jeff Norris, Senior Vice President of Finance. Please go ahead.

Jeff Norris

Thanks very much, Josh, and welcome, everybody. To access the live webcast of this call, please go to the Investors section of Capital One's website, capitalone.com. A copy of the earnings presentation, press release and financial supplement can also be found in the Investors section of Capital One's website, capitalone.com, by selecting financials and then quarterly earnings releases.

With me tonight are Mr. Richard Fairbank, Capital One's Chairman and Chief Executive Officer; and Mr. Andrew Young, Capital One's Chief Financial Officer. Rich and Andrew are going to walk you through the presentation summarizing our third quarter results for 2026.

Please note that this presentation may contain forward-looking statements. Information regarding Capital One's financial performance and any forward-looking statements contained in today's discussion and the materials speak only as of the particular date or dates indicated in the materials. And Capital One does not undertake any obligation to update or revise any of this information whether as a result of new information, future events or otherwise. Numerous factors could cause our actual results to differ materially from those described in forward-looking statements. For more information on these factors, please see the section titled Forward-Looking Statements in the earnings release presentation and the Risk Factors section of our annual and quarterly reports accessible at Capital One's website and filed with the SEC.

And now I'll turn the call over to Mr. Young. Andrew?

Andrew Young

Thanks, Jeff, and good afternoon, everyone. I will start on Slide 3 of tonight's presentation. In the third quarter, Capital One earned $3.5 billion or $5.61 per diluted common share. The results for the quarter included several adjusting items related to the Discover and Brex acquisitions, which are outlined on the slide. Net of these adjusting items, third quarter earnings per share were $6.69.

Relative to the prior quarter, both period end and average loans held for investment increased 1% while period end and average deposits decreased 1%. Linked quarter revenue increased 3% and noninterest expense declined 2%, resulting in pre-provision earnings growth of 10%. On an adjusted basis, pre-provision earnings grew 9% quarter-over-quarter. Our provision for credit loss decreased $73 million or 2% to $2.9 billion in the quarter. The provision reflects $3.1 billion of net charge-offs and an allowance release of $225 million.

Turning to Slide 4. I'll cover the allowance in greater detail. The $225 million allowance release in the quarter brought the allowance balance to $22.7 billion. Our total portfolio coverage ratio decreased 10 basis points and now stands at 4.92%. I'll cover the drivers of the changes in allowance and coverage ratio by segment on Slide 5.

In our Domestic Card business, we released $250 million of allowance. The allowance release was driven by continued favorable observed credit in the quarter, partially offset by greater consideration to downside economic scenarios related to geopolitical risks. The coverage ratio decreased by 13 basis points and now stands at 6.86%.

In our Consumer Banking segment, we built $41 million of allowance. The allowance build was primarily driven by strong growth in the auto business, partially offset by an improved outlook for vehicle values. The coverage ratio ended the quarter at 2.34%, 5 basis points lower than the second quarter.

And finally, in our Commercial Banking segment, the allowance balance was flat at about $1.5 billion, and the coverage ratio declined by 2 basis points to 1.6%.

Turning to Page 6. I'll now discuss liquidity. Total liquidity reserves ended the second quarter at about $134 billion, down $10 billion from the prior quarter. Our ending cash position decreased by about $6 billion to approximately $49 billion. The market value of our investment securities decreased by about $3 billion. Our preliminary average liquidity coverage ratio in the quarter was [ 153% ].

Turning to Page 7, I'll cover our net interest margin. Our third quarter net interest margin was 8.41%, 40 basis points higher than the prior quarter. The increase was driven by higher yield in the Domestic Card business, a $13 billion decline in average cash balances and an additional day in the quarter.

Turning to Slide 8. I will end by discussing our capital position. Our common equity Tier 1 capital ratio ended the quarter at 13.8%, roughly flat to the second quarter as net income in the quarter was largely offset by share repurchases of $2.6 billion dividends and growth.

With that, I will turn the call over to Rich. Rich?

Richard Fairbank

Thanks, Andrew, and welcome, everyone, to our earnings call this evening. Slide 10 shows third quarter results in our Credit Card business. Credit Card segment results are largely a function of our Domestic Card results and trends, which are shown on Slide 11. The Domestic Card business posted another quarter of top line growth and strong credit results. Year-over-year trends in domestic card results now fully incorporate the impacts of the Discover acquisition. Year-over-year and quarterly comparisons continue to reflect the much smaller effect of adding Brex to the Domestic Card business and moving our small legacy Corporate Credit Card business from the commercial bank to Domestic Card in the second quarter of this year.

Third quarter purchase volume grew 11% year-over-year, primarily driven by growth in legacy Capital One purchase volume. The addition of Brex and the Corporate Card business in the second quarter was a modest tailwind to purchase volume growth, while legacy Discover purchase volume was a modest headwind. Because of the Discover loan growth brownout, legacy Discover purchase volume was essentially flat compared to the prior year quarter.

Ending loan balances increased about 2% year-over-year. The legacy Discover card loans shrank about 3% from the prior year as the temporary brownout of Discover loan growth continued. Excluding Discover, ending loans grew [ about ] 6%. Revenue was up 5% from the third quarter of 2025, driven by growth in legacy Capital One purchase volume and loans. Revenue margin for the quarter was 17.8%. The Domestic Card charge-off rate for the third quarter was 3.98% down 73 basis points from the prior quarter and down 65 basis points year-over-year. Our charge-offs are benefiting from the steady improvement in delinquencies that we observed in the first half of this year and from strong recoveries performance. The delinquency rate was 3.74% at quarter end, up 35 basis points from the linked quarter and down 15 basis points from a year ago. The linked quarter trend was in line with normal seasonality.

Domestic Card noninterest expense was up 9% compared to the third quarter of 2025, driven by continuing technology investments, higher marketing and the addition of Brex. Our choices in Domestic Card are the biggest driver of total company marketing. Total company marketing expense in the quarter was about $1.6 billion, up 14% year-over-year, driven by higher direct marketing in our Domestic Card business, increased media spend and continuing investments in premium benefits. As we look ahead to the fourth quarter, we expect the Q3 to Q4 percentage growth in marketing to be similar to the Q3 to Q4 percentage growth we saw in 2025, primarily driven by seasonality.

Slide 12 shows third quarter results in our Consumer Banking business. Global payment network transaction volume for the quarter was approximately $191 billion. Network transaction volume increased 25% compared to the third quarter of 2025, driven primarily by the successful conversion of Capital One debit customers to the Discover Network. Auto originations were up 22% from the prior year quarter. We continue to be in a strong position to pursue resilient growth in the current marketplace.

Consumer Banking ending loan balances increased about $11 billion or about 13% year-over-year. Average loans were up 12%. Both ending and average deposit balances declined about 1% from the linked quarter, driven by volume impacts and choices related to the Discover integration including the sale of a small IRA portfolio we acquired from Discover. Legacy Capital One retail deposits continued to grow.

Consumer Banking revenue for the quarter was up about 17% year-over-year, driven predominantly by Discover revenue synergies and growth in auto loans. Noninterest expense was up about 10% compared to the third quarter of 2025, driven by increased auto originations and continued technology investments. The auto charge-off rate for the quarter was 1.59%, up 5 basis points year-over-year and up 16 basis points from the sequential quarter. The year-over-year increase is largely the result of a gradual mix shift in new originations and loans as our subprime mix is returning to pre-pandemic levels. The auto delinquency rate was 4.64%, up 32 basis points from the linked quarter and down 35 basis points from the prior year.

Slide 13 shows third quarter results for our Commercial Banking business. Compared to the linked quarter, both ending and average loan balances were up about 1%. Ending in average deposits were also up about 1% from the linked quarter. Revenue increased 6% from the linked quarter with strong growth in noninterest income. Noninterest expense was up about 1%. The Commercial Banking net charge-off rate for the second quarter was 0%. The commercial criticized performing loan rate was 4.4%, down 4 basis points compared to the linked quarter. The criticized nonperforming loan rate was down 10 basis points to 1.22%.

In closing, third quarter results continued to reflect solid top line growth and strong credit performance. The third quarter is a seasonally strong quarter in most years with seasonal strength in margins and charge-offs. And this year, the quarterly results also benefited from an allowance release.

Looking ahead, we expect to lean harder into marketing in 2027, driven by several factors. We're fully pursuing our origination opportunities and the window is open across our businesses. Our biggest competitors continue to invest heavily in marketing, and we need to be at the top of our game. Also, we plan to now lean back into Discover marketing. The trend of marketing increases is also powered by a gradual mix change some of the biggest growth areas in the company. Our businesses where marketing is inherently a much larger component of the business model. That includes our quest at the top of the market with heavy spenders, Capital One Shopping and our National Banking business. These businesses take a lot of marketing to build but are great franchises with lower risk, lower capital and long-term revenue annuities.

The Discover integration continues to go well, and we're making good progress on delivering the expected synergies. On a combined basis, we're achieving the expected synergies ahead of schedule. Our year-to-date results reflect about $1.8 billion of annualized combined synergies primarily driven by the successful debit conversion early in 2026 and the pace of achieving expense synergies. We have talked for a while now about the earnings power on the other side of integration. You can squint your eyes and see it now in the first 3 quarters of 2026. You would need to normalize for sizable -- for the sizable allowance release we've had thus far this year and to add the increased marketing we expect in the fourth quarter.

Looking forward, you would add in the remaining synergies we expect to get and probably deduct the lost interchange revenue that will likely come from the Visa, Mastercard settlement. Our current view is that the remaining synergies and the Visa, Mastercard settlement are roughly offsetting in magnitude. With those adjustments, that's a view into the run rate earnings power coming out of this deal, and it's consistent with what we expected at the time we announced the deal. Pulling way up, we've worked for years to put ourselves in a strong position to win in a rapidly changing marketplace. We continue to lean into investments in modern technology, data and AI. We continue to gain traction as we build enduring franchises with heavy spenders at the top of the credit card marketplace and with primary banking relationships, in our national retail bank. And we're building momentum from the game-changing acquisition of Discover.

And now we'll be happy to answer your questions. Jeff?

Jeff Norris

Thank you, Rich. We'll now start Q&A session. [Operator Instructions] Josh, please start the Q&A.

Operator

[Operator Instructions] Our first question comes from Terry Ma with Barclays.

질의응답

Terry Ma

I appreciate the color on the fourth quarter marketing and also the outlook for marketing in 2027. But as we kind of think about it, how do you measure the return on the investment for those marketing dollars? Maybe for example, the dollars allocated to the Discover brown out since there is a lag between deployment of those dollars to account generation to balance growth, like should we expect sort to keep leaning into marketing ahead of the loan growth materializing?

Richard Fairbank

Thank you, Terry. So we built a company that was designed for organic growth. Ironic that I say this at the time, we're talking about having done the big acquisition of Discover. But marketing is right at the heart of this model. And as we've talked about many times, we run our company based on horizontal economics, where before an investment before, during and after an investment, we horizontally measure the performance and make adjustments if necessary. But the key thing is that methodology is really quite different from vertical accounting where one is trying to see how anything is doing based on vertical performance.

Vertical-focused systematically underweights and underappreciates the value of investment over the longer term. So that's why if you pull way up on Capital One with our philosophy of horizontal accounting and building an organic growth company, we have, over the years, invested heavily to be where we are today. And over time, the vertical returns in some sense, on a lag basis manifest that.

Let's talk about marketing in particular. So for starters, the marketing machine of Capital One, much of it is a direct marketing machine that is pretty much stimulus response measure things pretty quickly with respect to what are we -- what is -- what we're getting in terms of booking new accounts for an investment in marketing. But there's also a whole other component that doesn't lend itself as easily to stimulus response thing, and I put at the top of that list building a brand. And -- but also included in that is it's hard to measure the return on investments in some of the things we're doing at the very top of the market, the unique experiences, the lounges, the -- so there are -- this will never lend itself to be all of it and exactly stimulus response measurable effect. But what we do is we pull up not only with respect to each choice that we make, but then pull up with respect to larger aggregations like our businesses themselves and look at their performance over time, and that's a very, very important part of our decision-making.

The other thing I want to say -- but if we stand back, it is, to your point, Terry, it's striking to notice that the growth rate of marketing over the last number of years has been higher than the growth rate of revenue, the growth rate of other expenses. And so that, of course one can stand back and ask, but wait a minute, that's what exactly is going on. And that's why I wanted to make the point here in my earlier remarks that the businesses where we are really leaning in and growing the most, our businesses that have a high marketing component to them. Obviously, the winning at the top of the market with heavy spenders is a very sort of marketing intensive business, high cost of acquisition and high cost of things that the customers care a lot about, the building a brand, all of that.

But I also want to point out on the National Bank. All the other major banks out there that the growth strategy they have is buying other banks. And that's not our growth strategy. We are organically building a national bank. And without a lot of branches on the corner, marketing is really the engine for that. So but my point was there is a gradual mix change going on where marketing is a higher component of our growth businesses and therefore, in some sense, becoming a higher component of our business. But through all, we continue to be as bullish about marketing and our -- the return on our marketing as we have been really at any time in this journey.

Terry Ma

Got it. And I guess just as a follow-up, you called out the Visa, Mastercard settlement. Any way to kind of quantify the impact to Capital One? And I guess you kind of tied it to the synergies, is there any more color that you can kind of give?

Richard Fairbank

So Terry, the revised settlement between merchants and Visa and Mastercard is going through the court approval process now. The hearing on final approval will take place in November. I know a lot of people in the investor marketplace or maybe even analysts are saying, "Well, we'll wait to see what happens and then sort of put it into our estimates." We put a pretty high likelihood on the final approval of this. And so in our planning, we have for some time been incorporating our best estimate that we can. This has a pretty wide error bar. We don't really have great insight into what that number is going to be, in many ways, we see the same information, the same analysis that you do.

So my point was just that we are planning for this to happen, maybe it won't, but we're planning for it to happen. And then my point also was that in sort of rough ballpark, that is -- if we sort of stand from where we are today, looking at that impact on run rate is very -- it is in the ballpark of the impact of the extra synergies that are coming our way beyond the annualized number that we mentioned before. So those are in a nice way sort of offsetting a nice way, meaning that for us to kind of give you a window to reach out and touch the earnings power. I think those things are generally offsetting.

Operator

Our next question comes from Ryan Nash with Goldman Sachs.

Ryan Nash

So Rich, maybe to pick up on your ending statements. There were a lot of moving pieces, cost saves, network settlement, reserve normalization and elevated marketing, I think, were the 4 things you highlighted, I guess, what are we supposed to take away from that statement? Are you saying that I tried to do math quickly, but return should be at 20% plus? Or is there something else underlying in the statement? And I guess given your response to Terry's question, should we expect that you'll be maintaining these returns? Or do you think marketing will eat into some of them? And I have a follow-up.

Richard Fairbank

So thank you for your question, Ryan. So let's pull way up on the guidance that we've been giving over this since we brought Discover on because we don't generally guide on things like specific earnings power, things like that. But I think the question sort of on everybody's mind has been this kind of amazing deal with Discover. What's going to be the earnings power coming out the other side? And how will we -- when can we sort of validate that we see that?

And so what we've been saying along the way is there are a number of things that have already sort of moved but that we still believe that the earnings power coming out the other side of this integration is consistent with what we believed at the time that we made our announcement. And what has happened along the way as we were getting ready for this earnings call and just savoring how much of the synergies have, in fact, been achieved that that's sort of the biggest thing that still as been waiting to happen as this deal is integrated.

And so our point was, given that everyone wants to see what is the earnings power. And how does Capital One view the earnings power coming out of this deal. My point was with a couple of approximate adjustments, we can actually just see it right now. We don't have to hold our breath and wait for it. It's really -- we can see it right now, and that's -- we don't we're not in the precise number business. This is -- as you know, in our business and earnings numbers involve the subtraction of very large numbers. They involve credit numbers that can move around and allowance numbers and things like that. But I think that what we have been saying for what we think will come out the other side of this deal is, in fact, manifest right now with a small number of basic adjustments. And that's really our point and we celebrate that it's there to be seen and that's really our -- the point that we wanted to make to investors.

Ryan Nash

Got it. And maybe, Rich, just as my follow-up. Obviously, there's a lot of macro noise out there, but the credit performance remains very strong. Just maybe unpack for us what's driving the improvement in card? And as you look forward, do you think that we can continue to see this type of improvement given the dynamics you're seeing in the portfolio?

Richard Fairbank

Okay. The -- let's just talk about our credit performance. I mean, we've got I think Capital One is performing very well in credit and there's kind of an industry effect there. And I think even on top of it, there's some Capital One very positive things going on. But the -- but if we pull up on credit for a long -- for quite a long time now that extends back into last year. Credit was -- has been improving relative -- a little bit kind of every month or every quarter to coming in just a little bit better than seasonality. So there's been this cumulatively very significant improvement that has happened. I'm talking about delinquencies now.

In the last few months, I think what we see is more in line with seasonality rather than outperforming seasonality. But we're in a very good place with respect to credit. Now if you think about charge-offs, there are a couple of things to say there. First of all, charge-offs are lagged relative to delinquencies. The charge-offs continue to improve as the delinquencies have improved. And then we have the striking recoveries effect that I just want to sort of reflect on that. Recovery is, of course, an important factor in our overall credit performance. And our recovery rates have historically been at the high end of the industry. probably as a result of our business mix and our strategy. And during the period of post-pandemic normalization, we noted that our recoveries were unusually low because of record low charge-offs during the preceding 3 years. And those low recoveries were a factor pushing overall loss rates well above pre-pandemic levels.

Today, we're seeing the opposite effect. Our recoveries inventories are higher than normal because of higher charge-offs in the rearview mirror. This is even more true at Discover, which had significantly higher charge-offs 2 years ago than they do today. But even when we adjust for this inventory effect, we see stronger recovery rates on recent charge-offs than what we've observed historically. And we think at least some of that is driven by sustainable improvements in our recovery strategies enabled by our transformation. So pulling way up on credit we think credits in a solid -- in a very good place. We'll note that delinquencies have been kind of moving with seasonality for a few months now as opposed to beating it. But all told, we see it in a very good place.

Operator

Our next question comes from Sanjay Sakhrani with KBW.

Sanjay Sakhrani

Quick question on expenses again, Rich. When we think about sort of the direction of travel on the efficiency ratio from here, how should we think about it? Should it be improving from here on out? And then just on the settlement, I guess, yes, we can -- it's a subjective thing. It may get appealed and delayed, and I get your point. But are there any offset to that interchange reduction that you could think of? Like would you pull down the awards to offset some of that impact? I'm just trying to think about the net effect of the decline in interchange part for the settlement.

Richard Fairbank

Okay. Thank you, Sanjay. So let's talk about the efficiency ratio. There's a lot of variability quarter-to-quarter in efficiency ratio. So of course, I wouldn't anchor to any ones quarter results. And the efficiency ratio will continue to reflect the realization of synergies and our investment imperatives. And we continue to lean into our investment imperative, including foundational technology, AI, as well as important growth factors like heavy spenders and our National Bank and our Global Network, and we will step up marketing investment.

And as I mentioned also, our biggest growth areas, our business is where marketing costs are a bigger component of the business model and which have inherently higher efficiency ratios. But also, of course, they tend to bring lower risk, lower capital and long-lasting annuities. So there's a little bit of a geography change across a lot of the line items of Capital One more gradual from quarter-to-quarter. But over the longer arc of time can be important. But also just we continue to lean in very hard to harvest meaningful efficiencies outside of our investment areas powered by our technology transformation. So there are a lot of moving pieces. Some of them are structural. We're not giving specific guidance on efficiency ratio, and we've been more focused in our dialogue on earnings power.

And -- but we -- there really is the sort of continuing efficiency improvement underneath Capital One and so much of the business that we run while on the other hand, there's a mix change and a real leaning into investments that make the netting of these sometimes move in opposite directions. So -- or the netting of these 2 -- just to lead us not to give specific guidance on efficiency ratio.

With respect to the settlement, so in some ways, the settlement is a couple of decades in the making. So I think folks could say, we've seen this extended so many times like why are we talking about it now. As I said earlier, from what we can gather, we give reasonably good odds that the settlement will happen. But the way we manage the company, when costs are likely to be coming, we put them into our numbers and manage accordingly. So we -- so from an expected value kind of point of view, that we're absolutely managing to that.

The -- there the specific -- that specific change, I don't think is necessarily will catalyze any one specific change tactically in what we have because we've been incorporating the expectation of that for a long time in the strategy and the rewards that we set, et cetera.

Operator

Our next question comes from Rick Shane with JPMorgan.

Richard Shane

Rich, I'm really curious about the Brex acquisition. When we think about Discover, you acquired a network, but I don't think you would argue that the technology there was cutting edge. Brex obviously has had a very different approach to market, brand-new platform. I'm curious how much of the technology from Brex Capital One is actually integrating and adopting and changing your workflows. Is there learning from them?

Richard Fairbank

So Brex is an amazing company. You -- I think you all saw my excitement at the day of the announcement and we have had it ever since. But I want to -- there are many things that are remarkable about Brex. But I do want to say that way up there, one of the most remarkable things is that this -- most founder companies these days are very tech forward. And just in terms of [ PureTech ] shops, innovation, they're way up there right at the top. But in addition, the choices that they made amaze me. And I have set reference for it because I think we make a lot of similar choices in the building of Capital One, and that was choices to build for the long term, build from the bottom of the tech stack and be very referential about things that will be profoundly helpful later, but that cost more money now.

And so the fact that they have a full modern tech stack that they built themselves for their the age that, that company is breathtaking and it's really valuable will be very valuable to Capital One over time. We have a great time hanging out with the Brex people comparing tech practices learning and there's so much to learn from them. And then I think that they've also seen a lot about how a company like Capital One which has been built as a tech company has scaled technology and also taking on things like how do you create a scalable risk management in a highly regulated environment and things like that. So there is a lot of learning.

I think one of the areas that Brex I think most commonly struck by is where Capital One is on the data and analytics side of the business. Of course, we built the company on that. And so there's just so much I think that we are sharing, and we look forward to share.

So let's talk about integration of technology. We are taking a -- we see great upside down the road to integrate a bunch of things, but we're also being very targeted and measured in our approach to integrating our technology with Brex, so we can be sure to allow them to focus on growing their business. And another striking indicator of the quality of what they've built is that we can let whole parts of Brex sort of operate independently and be confident about that because of their own reference for risk management and how they build things. So as we've contemplated the broader integration, we will slope the work focusing it, and we already are doing this, focusing on the critical risk management requirements that come with being part of a regulated national bank.

Now over time, Capital One can leverage Brex's spend management tools to broaden and enhance our offerings for our existing small business personal liability card customers, we are wildly excited about that opportunity. But that's not something we can sort of reach over and grab that. That's going to require more connection and more integration of certain tech and data aspects of this. So that one's down the road. We will also integrate Brex's spend management platform with the Capital One travel portal which will allow businesses to manage their corporate travel expenditures and policies, driving growth in business travel revenue. And as a side point there, we have really been leaning into our own travel business. I would encourage folks who are cardholders to check out our stand-alone travel app that we have and really quite a lot of great things there, and we're building many more things on that.

But anyway, so we look forward to that synergy -- strategic synergy that will come there.

And lastly, we will integrate Brex's banking services with ours, which will provide us capabilities to leverage Brex's business banking software with our payment capabilities. So this is -- so picture this sort of integration phase is the first phase is a little bit more which I colloquially call adding water that we try to do the things, the most important integrations and then add water to help grow their business. And Brex -- and just comment on that for a sec is already experiencing some of the early tailwinds that come with our brand, and we expect these to only grow stronger over the next few months. They're benefiting from the cost of funds, from moving to our balance sheet. We have already stood up a program to share high potential leads from across our businesses with Brex and are seeing very promising results at the outset, and we will scale into this approach more completely over time.

We -- over the coming months, as we test and learn, we'll start leaning in with our marketing dollars. And what's also coming in this sort of ad water phase but not yet at this moment is fully leveraging the marketing machine of Capital One, which requires a little more technical integration, setting up data pipelines calibrating our models for Brex's customer base and so on. And for our travel business, we're going to be focused on the hopper build out through the balance of the year. So the travel portal -- bringing the travel portal to Brex is probably be a next year kind of thing. But anyway, that's a window into how we view this acquisition.

But I just want to favor the -- that one of the power benefits of a deal like this is taking an absolutely leading technology company that is transforming a financial service marketplace and bring it on to Capital One with our modern tech stack and heritage of building a tech company. And I think that we're going to be able to create a lot of opportunity, and it's an opportunity that can uniquely be made possible by this combination.

Operator

Our next question comes from John Pancari with Evercore.

John Pancari

Rich, I appreciate you walking through the process of how you got to the earnings power and how we should think about that. So given the way that you framed it in that earnings power is in sight, is it fair to say that based on that, we should not expect a meaningful incremental step-up in expenses from here as you continue with your business and network investments and that this ballpark of where we triangulate given the math you gave us that seems like it may be in that 20% ROCE level that, that does not -- that we should not dial in an incremental step up in expenses as we look forward?

Richard Fairbank

Well, John, there are a lot of moving pieces in how we think about this company and where we go with our strategy. My real point was the earnings power that you can see out of this combination is now you can pretty much reach out and touch it. We have been saying along the way, we are leaning hard into investments. And a lot of that leaning is already in the numbers that we are printing these days. But there is important leaning that's coming going forward. And the -- and I commented things like the top of the market, I'm not going to run through the whole list necessarily again.

But we will continue to lean into investments. And our belief is that we have the earnings power to be able to be able to sustain pretty high levels of investment to help a really nice growth future and, at the same time, deliver strong earnings power, reflective of the kind of thing that you see.

John Pancari

Okay. Got it. And then separately, just around the network. Can you just update us on the progress of moving some of the Capital One Credit Card volumes onto the Discover Network and any the pilots that you're running? How should we think about the pace of that migration in the time frame involved?

Richard Fairbank

Okay. Thank you, John. So as I often say, the Discover Network is a rare and valuable asset and the crown jewel of the acquisition. Discover built an amazing network. To this day, the deeper I dive into networks it still amazes me how far they came with building out a global network with the scale that they had. They were benefited by being an early player going back decades and -- but I give so much credit to them, but they didn't have the scale to fully complete and sort of turn the flywheel. So they were trapped in a subscale situation doing an amazing job with what they had.

The Capital One brings the opportunity to build more scale in a very scale-sensitive business. I don't know if I've seen a more scale-sensitive business than this one. So earlier this year, of course, we completed the conversion of our Debit Card business to the Discover Network. And we are very pleased with how the conversion went from a customer experience standpoint, the volume retention and the growth that we're driving on the other side. We're also looking forward to growing credit card volume on the Discover Network in several ways. So we will, of course, continue to lean in to grow Discover branded cards, which have always been on the Discover Network and will continue to be.

We will also build network volume by originating Capital One cards on the Discover Network and by converting Capital One customers from their existing network to the Discover Network. And we've been conducting test to help inform our strategy regarding which customers and products we target for moving to the Discover network and the timing of those moves. In parallel, we will lean hard into enhancing global network acceptance and building the Discover brand. And along the way, we will slope the work as we drive acceptance and slope the work in selecting which products and customers are moved onto the network.

So this will be a -- this is not going to be a big bang thing like you saw on the debit side. But this is -- we want to make very clear. This is what we're working backwards from and picture us leaning in, really giving it the best opportunity it can, investing to be able to get there and then sloping the work on the acceptance and the conversion side along the way.

Operator

Our next question comes from Andrew Bauch with BMO Capital Markets.

Unknown Analyst

This is [ Conan ] on for Andrew. My question is on the premium card space and Venture X specifically. I was wondering if you can talk about the competitive landscape and how you view the lifetime value of an incremental premium card user over time, just given the increase in rewards and marketing spending.

Richard Fairbank

So thank you. So the launch of Venture X is a continuation of our journey over many years to win at the top of the market. And it really started back in 2010 when we launched our Venture Card, and we have continued, it's another one of these things, the Capital One, where we declare where we're going and patiently build something over an extended period of time, which can spend. In this case, we're well into our second decade of this quest.

But included in this has been the launch of our Capital One travel portal, the opening of Capital One lounges and investments in sort of groundbreaking customer experiences and exclusive services. We have been very pleased with the customer engagement and the traction that we're getting. And we have with each -- every year or two, we just keep moving a little bit higher in the marketplace stretch higher toward the heaviest spenders. And over this period of time, we continue to see that the fastest growing part of our card business is with the heavier spenders. And I think there's so much growth opportunity left there.

But first -- but the thing we note is the very positive growth trajectory that we get and the very positive customer validation that we get. When we think about economics, we let's go back to our journey from 2010. We set out on this quest, this battle cry, we're going to win at the top of the market. And we had Venture and [ Venture One ], our first 2 products, and we lean into those, build them to scale look at the very impressive horizontal returns coming from that franchise. We lean up. We go into Venture X and we watch how that thing is going and the -- and the power of the earnings power of those annuities. Meanwhile, on the small business side, we have had a parallel quest going on with our Venture X business card. We're very pleased with those results.

So we -- I guess with -- so we validate along the way with our horizontal economics, the customer the traction with customers, the growth rate, the annuities, the attrition rates, the charge-off rates that all collectively add up to the overall economics. But what we continue to see is a very strong validation of this Quest. And -- but included in it, we know that it involves continuing to invest grow not only more customers, but to keep growing higher in that marketplace. So we're pretty excited about where we are.

Just a comment, you asked about the competitive -- maybe just a comment about the competitive position of, for example, Venture X. We feel we're in a great place with Venture X's competitive positioning as a straightforward high-value product. And we see our competitors doing leaning in very hard. And I think everyone is taking out a bit of a special place in that market at the top of the market. But I think the power metrics that we're seeing from our quest are very impressive.

Operator

Our next question comes from Erika Najarian with UBS.

L. Erika Penala

Just my first question, Andrew, on the net interest margin trajectory from the [ $841 million ]. Obviously, the revenue suppression being lower helped the earning asset yields. Lower deposit balances help with deposit costs. As we think about sort of growth in balances in the fourth quarter and maybe the puts and takes of how to think about the forward curve, can you sort of help us think about the path of net interest margin at least in the fourth quarter?

Andrew Young

Sure, Erika. Thanks for the question. The one thing I would add is just the benefit in the third quarter, too, is just the ending cash balances coming down in the quarter or sorry, the average coming down in the quarter was beneficial this quarter as well.

Looking ahead, something I noted last quarter is that all of the structural impacts from the Discover acquisition have largely played out at this point in our NIM. And so from a rate perspective, our NII is relatively neutral to rates, the pace and timing of rate hikes could impact NIM in any given quarter, just given the timing of asset repricing and deposit repricing decisions. But over, call it, a handful of quarters, the changes in short-term rates should have a largely neutral impact.

And so I would say longer term, the changes to NIM will be driven by 2 things. The first is the seasonality that happens by quarter, things like day count, revolve rates, you brought up the fourth quarter holiday spending impacting the denominator. Those are things that are going to impact quarterly NIM, but they're just not going to impact the ambient level.

The other thing ultimately that will impact NIM over time is just the relative growth rates of specific assets and liabilities that will be driven by business choices by the competitive environment by customer behavior. But at this point, the thing I would reinforce is the structural impacts to NIM have run their course. So it's more going to be about seasonality and the effects that I just described.

Operator

Our next question comes from Don Fandetti with Wells Fargo.

Donald Fandetti

Rich, recently Capital One ranked pretty well on an AI industry report. I guess I have 2 questions. One, are you seeing specific like revenue benefits, market share type gains from these investments? And number two, do you plan to quantify these benefits?

Richard Fairbank

So Don, the -- we certainly do not intend to quantify the benefits. But I want to stand back and just talk a little bit about AI and how we think about it and how central it is to who we are and what we're building. So the pace of advancement in AI is just breathtaking. The world is really, I think, never seen anything like the rate of innovation and change we're now seeing in the world of AI. The AI itself and then generative AI in 2022, agentic AI over the last year, it's just unbelievable and the pace of change is accelerating. And it is as clear as could be possible, at least in our mind that everything is going to change in our lives is going to change. And in business across the business world and in banking, it's never going to be the same. But those changes also aren't going to happen overnight.

But let me talk about that -- we've talked for a long time about our technology transformation. And you all know that in 2013, we pulled our company together and said, look, we built -- we set out to build a tech company and a data company, and I'm proud of what we built. But we're on the wrong side of history with respect to the technology that was modern in the past, but it's not modern now. And we set out to rebuild our technology from the bottom of the tech stack up. But I want to savor what the objective function of that rebuild was. Back then, we're talking more than a decade ago.

The objective function was not so that we could be modern. The objective function of the rebuild was AI and envisioning how the whole world was hurdling toward a world where everything was going to be about AI in real time, bringing instant solutions entirely customized to consumers and businesses and also creating the opportunity to transform how companies work on the inside. So that's what we set out in 2013. It's been the objective function ever since.

Now note, we didn't predict generative AI, and we did not predict agentic AI, but yet AI was the objective function of this transformation. And as I look at the amazing advancements in generative AI and agentic AI. They are part of a continuous curve of that is and the tech stack that it needs, we often say, well, if we had known about generative AI or agentic AI , what would we have done differently with what we build? And the answer is we would have built exactly what we built because we were working backwards from having a modern tech company were powered by AI.

So let's talk about the AI that's coming to the world, I think AI is everybody all over the place in business is talking about AI as they should be. And I think companies will adopt AI and get a lot of benefits. This includes banks and as well as other companies. But I think there will be the degree of benefit and the extent to which this is adopted will be across a really wide spectrum. And that spectrum relates to whether AI is we bring whether companies bring AI to improve work as it is done today or whether companies reinvent how work is done with AI. And I think everybody is going to do the former and get a lot of benefits in order to do the latter to reinvent how work is done with AI. It's I think it's going to need modern tech stack. It's going to need an organization that can work backwards from transformations that transcend business units and sort of collided with a lot of things about how companies are built and organized today.

So -- but we are bullish about these opportunities. And we -- when we think about -- so the customer experience, well, certainly, I think AI will be transformative for the customer experience because customized solutions in real time are going to be coming out the other side. But when you think about inside a bank, the opportunity with modern technology data and AI to transform the entire data ecosystem, how software is developed, credit, marketing, risk management, fraud, operations, servicing, cybersecurity pretty much I think, across the board.

And so these will be things that will -- and I think investors will often ask, well, where do I see AI? I want to see the app. I want to see it in the app or what's the manifestation and I appreciate those will be questions people ask but it's a little bit like asking the question, "Well, Capital One, you set out to build an information-based strategy 3.5 decades ago, where do I see it?" And my answer is it's all around us. It's everywhere, but I can't point it an app where you see it. It's just really in the business model in the choices in the way risk is managed customer experiences, marketing and everything.

And I think AI will be experienced by our investors in exactly the same way. And we are really bullish about the opportunity. But it also ties back to the question often investors are asking, why do we keep talking about -- why do we talk about investment? Can't we just settle back and just stay where we are with respect to our investment levels. You should all know -- I mean my view is we are experiencing something basically like fire and electricity itself, except experiencing it, not in the long period of time that those innovations came but in an incredibly short period of time. And we believe we're very well positioned, but we know there is a big investment imperative to take advantage of this. And it would be, I think, really a disservice to our investors if on the shoulders of what we have built, we held back in terms of investing in this extraordinary humanity changing thing that's happening.

Operator

Our next question comes from Bill Carcache with Piper Sandler.

Bill Carcache

As investors debate where we are in the cycle and how much growth runway is left? Many believe the Capital One franchise today is structurally more resilient given the Discover combination and the meaningful investments you've made in the digitization of the business and this quarter's results certainly support that. So while the Capital One in the past has historically gotten down to tangible book value and downturns, it would be helpful to hear how you think the resilience of the business has changed and whether you believe it's the business is in a superior position today to weather the next turn in credit, whenever that comes, of course, it's very favorable today, but it would be great to get your thoughts.

Richard Fairbank

Well, Bill, thank you so much for that question. Wow, there's a lot packed into this that you know from the founding of this company and maybe it's because I spent too many years consulting to companies that didn't -- that we're living the consequences of not doing what I'm about to say. But from the founding of this company, we have been just very focused on building something great for the long term. And that means something that not only has nice horizontal economics like we talk about, but something that has resiliency to weather the extraordinary storms that hit companies over time. And so we built our company with that philosophy and with respect to resilience, let's talk about what we've done. Always in our underwriting we don't underwrite to how things like the test cells are performing. We underwrite to a worsening scenario and assume the worsening is business as usual that we try to bake resilience into our very choices. And we have various resilience metrics we do.

We also take every program that we have at Capital One and evaluate it as part of this horizontal evaluation but evaluate the resilience of all the different parts of our business. And so it doesn't mean there's a guarantee of being resilient, but to bake worsening into the business model at all times is, I think, a really important thing. We also have a conservative capital philosophy. We've talked about that. And I think that we've been well served by that. But I believe that Capital carries -- Capital is very asymmetric risks and benefits associated with it, and we take a pretty conservative philosophy there.

Let's talk, Bill, about what has changed since the global financial crisis. Certainly, Capital One's business mix, we have just a lower-risk business mix in terms of the sort of credit content of the company way more spender led vendor-led portfolio, lots of business at the top of the market with heavy spenders, and I think that will serve us well.

We also, over the years, you may remember, Bill, I'm sure you do. We were vocal for years about managing to avoid the best we could, what we call high balance revolvers that we said they performed incredibly well in the good times, but the issue is resilience. And so we -- at the expense of growth over the years, really managed to do the best we could to minimize high-balance revolvers. By the way, Discover has got more of those than we do. And that's part of the trimming around the edges we've done there that has contributed to the brown out, but they have a great franchise.

So we -- those are some of the -- then along the way, also the technology of Capital One that we have built allows us to measure things so much better to monitor things in real time to monitor with even in some of our areas, you use AI-based monitoring that's that is full file instead of sample based, it's full file instead of batch it's real time instead of manual, it's automated. And instead of just flagging something, it's flagging things with root cause diagnosis and suggested solution. So the early days still on some of that AI monitoring, but that's just an example of where the technology that we have built at Capital One because from its founding day, we built basically a credit-based company. At the top of a list of what we build is to build the technology that can enhance our resilience. And so those are some examples.

I do want to say that one thing that was very wind aided in the global financial crisis that we should point out, and that is that we had a significant amount of our balance sheet of our loans off balance sheet. And so -- and also, it was not a CECL regime. So we -- when the sort of the bad times came. Our allowance build was way less than it would be pound for pound now with everything on balance sheet and everything with CECL and all that.

So -- but -- so I think there are some things a little bit on the other side that we should keep in mind. But one of the very strong motivators of the strategic journey of Capital One in the last decade has been the diversification of our business model and moving into businesses that have a lower -- lower credit risk, higher expense bases, lower capital requirements and that migration will continue.

Operator

Our next question comes from Darrin Peller with Wolfe Research.

Darrin Peller

Can we just -- I'd love to hear your opinion on the potential just given the political and the upcoming elections for [ CCCA ] and how you believe you could fare just given the obvious the Discover Network now, what that can mean for the business, both ways, whether it's gaining share or obviously, pricing dynamics of keeping things in-house?

And then just also on the topic of Discover, just as you were restarting marketing and originations around it, do you have a sense on the timing before we can see Discover return to positive loan growth? Any way to think about the economics of new balances there versus prior?

Richard Fairbank

Yes. So let's talk about [ CCCA ]. So interchange has been around for a long time because it's part of a well-established payments ecosystem in the United States that benefits both retailers and consumers alike. And we believe that there is substantial competition in the payments industry already with banks like Capital One taking a big portion of its interchange revenue and giving it to the consumer in the form of rewards. And in fact, if you -- when we have looked at the evolution of the international credit-based marketplaces. We find an observation is the banks in those markets probably passed a significantly lower proportion of their interchange to consumers.

I'm not faulting them, it's a choice that was made. The banks -- the leading banks in the United States, one of them being Capital One, but others as well have really leaned into taking such -- a big part of the interchange revenue and giving it to the consumer in the form of rewards. And and lots of important things happen. First of all, obviously, it's a big win for consumers. It stimulates consumer spending, merchant sales it helps the ecosystem thrive. And it also creates an important in a sense, political constituency that appreciates the power of what's happening there with consumers across the credit spectrum being able to enjoy their rewards.

So I think whether I speak from the legacy Capital One perspective or the Discover perspective, I think the ecosystem works very well now and a change that would inevitably alter credit card rewards would be undesirable for consumers, merchants and the economy. And so we -- that's a view that we have had, and it's a view that we still have even as we own this network. [indiscernible] the brown out excuse me. I get so excited about the other topic. Yes let me just talk about the brown up for a second.

So as we've talked about for a whole bunch of quarters now, the Discover Card portfolio is going through a bit of a loan growth, and I [indiscernible] created the phrase brown out just -- so it would have a branded term there. as several factors combined to pressure loan growth in the near term. Following Discover's credit expansion in their card business in '22 and '23, they dialed back their origination programs and credit line management by a fair amount toward the end of 2023, and they largely sustain those dial backs. Since we took over, we've been trimming on the margins of Discover's credit policies in areas where we are less comfortable with the resiliency of the underlying customers like high balance revolvers.

As a result of these pullbacks, the portfolio has been contracting and continues to face some headwinds to growth as these more recent smaller vintages mature. In the third quarter, Discover Card outstandings were down 2.9% year-over-year. By the way, it's worth noting the flip side of these pullbacks and the brown out has beaten strong credit performance. Now this brown out will not last forever since we will bring to bear a number of capabilities as we move Discover originations and existing customers to Capital One's technology. And that whole movement is way underway.

Getting Discover on to Capital One's technology will allow us to unleash our models, full-spectrum underwriting and spender capabilities to power more originations, higher spend volume and ultimately, higher loan volume over time. As of the end of September, 100% of Discover Card originations are on our tech stack. We are in the middle of our conversion of the back book, and we'll complete that in the first quarter. We are just one of our waves. We have several waves of conversion, but one wave happened over the weekend. You might want to peek at the App Store. Any of these days. I won't give it away, but check out the free apps on the App Store. And you'll see the evidence of droves of people moving to the Capital One mobile app.

So anyway, all these moves will bring Octane back to the portfolio. This quarter, new originations were up year-over-year for the first time since the deal closed. As the back book converts, we are resuming and leaning into customer management programs, which have been limited during the conversion. All these moves, of course, take time to turn into loans. We expect continued contraction over the next several quarters and then for loan growth to turn positive after that.

And just one -- I'm sorry, one other note, just the other area of brown out has been personal loans, and that's going through its own sort of magnified brown out because the -- those loans have shorter duration to them. And so the dial backs have been more dramatic. But over time, we have the same views for that one, and we look forward to leaning back into that.

Operator

Our next question comes from Rob Wildhack with Autonomous Research.

Robert Wildhack

Just a quick follow-up on the NIM commentary. Andrew, you teased it a little bit, but end-of-period cash was much lower than average cash in the quarter. So I'm wondering if there's any structural or longer-term opportunity to continue mixing away from cash? Or is this kind of it? And then just quickly, how about your thoughts on deposit growth in betas and deposit competition through the hiking cycle? There have been some mixed messages so far from some of the other players in that market. So just checking if there's any color you want to share there?

Andrew Young

Yes, sure, Rob. We ended the quarter with just under $50 billion of cash. The third quarter average was about [ 55 ]which is in line with where we ended the second quarter. Cash levels can move up and down due to the timing of deposit and loan growth has a little bit of seasonality in it and also need to keep an eye on the timing of maturities and wholesale funding issuances. But I don't want to give a specific target given the fluctuation. But I would say the range of this quarter sort of average and ending probably gives you a reasonable benchmark of what will be typical levels.

In terms of deposits, what we saw this quarter, as Rich noted in his talking points was deposits we're down about 1%. We sold a small IRA portfolio as part of the Discover integration that drove a majority of the decline this quarter, if you take a longer kind of aperture even with that decline, our book has grown something like 3.5% over the course of the last year. So we feel really good about the business we've built and the ability that it creates for us to have resilient high insurance percentage with deep customer relationship business.

So with respect then to beta, I would say the recent falling cycle had a lot of noise given we acquired Discover, which drove some mix change. So I don't believe that, that's particularly useful from an informational value perspective. But if you look at the last rising rate cycle that's probably a better neighborhood of what we could expect with this cycle's data, but it's important to note that each rate cycle is different. I think the ultimate beta is going to be driven by our product mix by the opportunities that we see to build those lasting customer relationships. The magnitude and pace of Fed funds move. But as you said, also the competitive landscape. And so keeping a close eye on that competitive landscape and how it potentially impacts pricing choices and beta, but we're starting from a place of believing that the last cycle is probably a reasonable starting point, at least for our assumptions, and we'll keep a close eye as this cycle plays out.

Operator

And our final question comes from Mihir Bhatia with Bank of America.

Mihir Bhatia

I just have one quick one. I just wanted to clarify the comments on earnings power in 2027 investments. I think Rich, you said the deal model earnings power is visible today, making a couple of adjustments from today's results. But then you also, I think, Q&A have talked about a lot of important investment that's still to come and that holding it back would be a disservice to investors and we totally understand you want to invest for the long term.

But I just want to be clear about the earnings power between as we think about 2027, 2028. When you talk about the earnings power that you have projected at the deal, does that include those step-ups in marketing, technology, spending? Or does that spending power come on top of the model, if you will, right? Like it's -- just trying to understand if it [indiscernible] incorporated some of this investment spending you're talking about.

Richard Fairbank

So the -- thank you for that question. The Investments. We have already been leaning into investments significantly in our company. We've talked a lot about that. And so the -- a lot is already in the numbers that you see and we are continuing to lean in as well, some things that are probably going to see more in the future than they have our investment in the Discover Network, which is really just kind of getting underway. We talked about the growth in marketing, of course. But -- and the investment in AI and capitalizing on this opportunity.

So I we aren't generally going to be in the business of being in the year after year giving guidance on the earnings power of the company. I think our -- really, our point has been we wanted to -- wanted to show the earnings power that's on the other side of this deal, and we have -- and I think you can -- our point is you don't have to hold your breath, you can reach out and touch it here. We will continue to lean into investments as we do at Capital One, and there's quite a bit that we are pursuing, but it's also the case that we are very focused on the earnings power in this company and generating quite a few savings on the -- on one minus those investments.

So we just wanted you to be able to reach out and see the earnings power as the dust settles from the from the integration. And we continue to be bullish about the earnings power going forward as well as leaning into important investment opportunities. And that's a little bit of a window into who we are.

And I want to reflect on that because from the founding of the company, we didn't build -- we didn't just say we're going to go do what all the other banks did. We very carefully chose businesses that had inherently high earnings power that enabled the delivery of strong earnings power and at the same time, investing to build a great company for the long term. And I think you're seeing it today and so I think it's simultaneously the case that we're -- we have a lot of earnings power on display, and we will continue to lean into very high-value investment opportunities. And always with -- and that I think that's going to be a key formula for driving great earnings power for years.

Jeff Norris

Thanks, Rich, and thanks, everybody, for joining us on the conference call this evening. Thank you for your interest in Capital One. Have a great night.

Operator

Thank you. This concludes today's conference call. Thank you for participating. You may now disconnect.